Trends in Underage Drinking in the United States, 1991–2023 | NIAAA

Chiung M. Chen, M.A.
Hillel R. Alpert, Sc.D.

CSR, Incorporated¹
Suite 400
20130 Lakeview Center Plaza
Ashburn, VA 20147

July 2026

U.S. Department of Health and Human Services
Public Health Service
National Institutes of Health

¹ CSR, Incorporated, operates the Alcohol Epidemiologic Data System (AEDS) under Contract No. 75N94023C00015 for the National Institute on Alcohol Abuse and Alcoholism (NIAAA). Dr. Bradley Kerridge (Division of Epidemiology and Prevention Research) serves as the NIAAA Contracting Officer’s Representative on the contract.

Highlights

This surveillance report, prepared by the Alcohol Epidemiologic Data System (AEDS), National Institute on Alcohol Abuse and Alcoholism (NIAAA), presents data on underage drinking among youth ages 12–20 for 1991–2023. This is the ninth of a series of reports to be published every few years on underage drinking and related attitudes and risk behaviors. Data for this series are compiled from three separate nationally representative surveys: the National Survey on Drug Use and Health (NSDUH), the Monitoring the Future (MTF) survey, and the Youth Risk Behavior Survey (YRBS). The following are highlights of trends from 1991 through 2023. Note that beginning with the 2015 NSDUH and 2017 YRBS, the threshold for female binge drinking was changed to consumption of 4 or more drinks (from 5 or more drinks) on an occasion or in a row in the past 30 days.

Prevalence of Use

• Although there are marked differences in absolute values of estimates and changes in survey design and mode of administration, all three survey data sources show an overall decline in the prevalence of past-30-day alcohol consumption between 1991 and 2023. In 2023, 14.1% of youth ages 12–20 reported consuming alcohol within the past 30 days (NSDUH). Among 12- to 17-year-olds, the prevalence remains elevated, at 6.9%, exceeding the Healthy People 2030 target of 6.3%.
• Overall, the rate of decline in the prevalence of past-30-day alcohol consumption has been greater in males than in females in the last few years. The sex-specific trends for ages 12–20 converged in 2015, and the prevalence for females (15.0%) continued to exceed that for males (13.2%) in 2023 (NSDUH).
• Throughout this period, rates of underage drinking remained highest among non-Hispanic whites, followed by Hispanics and non-Hispanic blacks. Rates were also higher among youth ages 12–20 not enrolled in school as compared with those enrolled in school (NSDUH), although rates among full-time or part-time college students ages 18–20 remained higher than among their same-age peers not enrolled in college (data not shown).

Drinking Patterns

• The mean age of initiation of drinking alcohol has increased from 13.9 years in 1991–1993 to 14.9 years in 2021–2023 (NSDUH). In addition, there has been a gradual decline in the proportion of youth reporting initiating drinking at age 12 or younger (NSDUH, YRBS).
• In 2021–2023, youth drinkers ages 12–20 reported drinking on an average of 4.2 days in the past 30 days. They consumed an average of 3.1 drinks on the days that they drank. Their average total volume of consumption was 15.9 drinks in the past 30 days (NSDUH). Over the course of the study period, male drinkers have generally maintained higher average frequency, quantity, and volume of consumption in the past 30 days than female drinkers.
• According to NSDUH, the overall prevalence of binge drinking increased between 1993 and 2004, from 12.1% to 19.6%, but has since declined, reaching 8.3% in 2023 and aligning with the Healthy People 2030 target of 8.4%. By contrast to the household-based survey (NSDUH), data from the secondary school-based surveys (MTF and YRBS) show an overall decline in binge drinking rates; the downward trends appear to have started in 1999 (MTF) and possibly as early as 1997 (YRBS). Persistent gaps in binge drinking prevalence between males and females were observed over time prior to 2016, but by 2023, these gaps have narrowed (MTF) or even reversed (NSDUH, YRBS), with the overall prevalence higher in females (8.9%) than males (7.7%) (NSDUH). Compared to traditional binge drinking (5+ drinks in a row), there has been a greater annual percentage decline (MTF) in extreme binge or high-intensity drinking (10+ or 15+ drinks in a row).

Alcohol-Related Attitudes

• The percentages of youth who strongly disapprove of others regularly consuming alcohol or binge drinking and who consider regular or binge drinking a great risk (MTF) show a declining trend during the 1990s, particularly in the early 1990s. The trend reversed in the 2000s, either gradually increasing or remaining relatively unchanged for more than a decade. However, the percentages began to decline in 2016, with a further drop in 2021–2022 during the COVID-19 pandemic, followed by a rebound afterward.

Alcohol-Related Risk Behaviors

• Between 1991 and 2023, trends from the YRBS show an overall decline in the prevalence of driving while under the influence of alcohol among secondary school youth, and similar downward trends are observed in the NSDUH data after 2002. The NSDUH data show consistently higher prevalence but a more precipitous decline from 2002–2004 to 2021–2023 among non-Hispanic whites (11.9% to 1.8%) compared to Hispanics (7.5% to 1.4%) and non-Hispanic blacks (4.6% to 1.0%) for both sexes.

Introduction

This surveillance report on underage drinking is one of a series of reports published to monitor trends in alcohol consumption and alcohol-related morbidity and mortality. These reports are prepared by the Alcohol Epidemiologic Data System (AEDS), and Division of Epidemiology and Prevention Research, National Institute on Alcohol Abuse and Alcoholism (NIAAA), and are intended to be useful to researchers, planners, policymakers, and other professionals interested in alcohol misuse by young people and its associated consequences. In 2007, the Surgeon General’s Call to Action to Prevent and Reduce Underage Drinking brought renewed focus to this issue. The Call to Action reviewed the risk factors and outcomes associated with underage drinking, presented a developmental framework for understanding and addressing underage drinking, and identified six goals for the Nation to address the problem of underage drinking. This surveillance report responds to Goal 5, “Work to improve public health surveillance on underage drinking and on population-based risk factors for this behavior” (U.S. Department of Health and Human Services 2007, p. 37). The data presented herein also are essential in assessing changes toward meeting one of the Nation’s health behavior goals to reduce misuse of drugs and alcohol (e.g., underage drinking) as stated in Healthy People 2030 (U.S. Department of Health and Human Services n.d.). This is the ninth surveillance report on underage drinking developed by NIAAA. AEDS will issue follow-up reports on this topic every few years.

Rates of alcohol consumption among underage youth are a cause of concern (National Research Council and Institute of Medicine 2004), despite a documented decline in youth drinking across almost all high-income countries since the early 2000s (Caluzzi et al. 2022; Vashishtha et al. 2021). This downward trend was especially pronounced during the COVID-19 pandemic, when social disruptions contributed to notable reductions in both alcohol use and binge drinking among adolescents (Brener et al. 2022; Miech et al. 2023; Hoots et al. 2023). In this surveillance report, 14.1 percent of 12- to 20-year-olds reported drinking alcohol in the past 30 days (2023 National Survey of Drug Use and Health [NSDUH]). This is a decrease from the rate of 33.4 percent three decades earlier (1991 National Household Survey on Drug Abuse [NHSDA]), although changes in survey methodology preclude a direct comparison of these two estimates. The downward trend in underage drinking is mirrored by a steep decline in the prevalence of driving under the influence of alcohol, from 7.4% in 1995–1997 to 1.5% in 2021–2023 (NSDUH). In the United States, alcohol consumption begins early, with a mean age of 14.9 years (2021–2023 NSDUH), and 13.3 percent of high school students reporting that they consumed their first drink of alcohol before age 13 (Centers for Disease Control and Prevention [CDC], n.d.). By the 12th grade, 10.2 percent of adolescents report binge drinking (consuming 5 or more drinks in a row within the past 2 weeks) (2023 Monitoring the Future [MTF]).

Underage drinking is associated with an array of safety, legal, academic, social, emotional, behavioral, and health problems (Committee on Substance Abuse 2010; de Goede et al. 2021; Masten et al. 2009; McCambridge et al. 2011; Office of Juvenile Justice and Delinquency Prevention 2012; Ryan et al. 2019; U.S. Department of Health and Human Services 2025). For example, alcohol use caused youth to behave in ways they later regretted, interfered with their ability to think clearly, or caused them to drive unsafely. Youth who drink and drive are at increased risk of being involved in accidents because of the combination of alcohol-related impairment and relative driving inexperience (Haegerich et al. 2016; Peck et al. 2008). Underage drinking is associated with not only driving after consuming alcohol but also riding with a peer who has consumed alcohol (Li et al. 2018; Miller et al. 2007; O’Malley and Johnston 2013; Terry-McElrath et al. 2014; Vaca et al. 2020). Frequent binge drinking and high alcohol intake among youth increase the risk of higher dropout rates and lower grade point averages at graduation (Hjarnaa et al. 2023). When youth consume alcohol, they are more likely to engage in risky sexual behaviors, such as having unprotected sexual intercourse, having multiple partners, and being drunk or high during intercourse (Brookmeyer and Henrich 2009; Chernick et al. 2020; Cho and Yang 2023; Nkansah-Amankra et al. 2011; Oshri et al. 2013; Scroggins and Shacham 2021; Seth et al. 2011; Storholm et al. 2018; Stueve and O’Donnell 2005; Sun et al. 2024). This puts them at risk of adverse consequences, such as sexual victimization (Basile et al. 2020; Champion et al. 2004; Thompson et al. 2012; Waterman et al. 2019), unwanted pregnancy (Salas-Wright et al. 2015), and sexually transmitted infections (Chung et al. 2017; Cook et al. 2002; Khan et al. 2012). Adolescents under the influence of alcohol or with a history of drinking alcohol are also more likely to engage in violent behavior, such as dating violence, physical fighting, and assault (Blitstein et al. 2005; Jones et al. 2020b; Kedia et al. 2021; Kodjo et al. 2004; Salas-Wright et al. 2016; Swahn et al. 2004, 2013; Waterman et al. 2019; Wells et al. 2004). Adolescent alcohol use disorder and alcohol intoxication while depressed increase the risk for suicidal behavior (Chatterji et al. 2004; Esposito-Smythers and Spirito 2004; Ganz and Sher 2009; McManama O’Brien et al. 2014; Schilling et al. 2009; Wolford-Clevenger and Cropsey 2020). Adolescent binge drinking poses both acute and chronic health risks, impairing brain development, organ function, and the immune system. It also significantly increases the likelihood of developing alcohol use disorders; sustaining risky drinking behaviors; and experiencing a wide range of mental, physical, and social problems in adulthood (Del Ciampo and Del Ciampo 2024; Wallace et al. 2024). Several studies have established a convincing link between youth alcohol use, particularly binge drinking, and illicit drug use (Esser et al. 2021; Jones et al. 2020a; Miller et al. 2007; Miller et al. 2024; Siliquini et al. 2012), and identified distinct brain networks that predispose adolescents to risky behaviors, such as experimenting with drugs and alcohol (Antón-Toro et al. 2022; Chen et al. 2025; Hammerslag and Gulley 2016; Jacobus and Tapert 2013; Lees et al. 2019; Squeglia and Cservenka 2017; Squeglia and Gray 2016; Whelan et al. 2012; Yip et al. 2023). In 2023, among youth ages 12 to 17 who were heavy drinkers (i.e., having consumed 5 or more drinks for males or 4 or more drinks for females on the same occasion on each of 5 or more days in the past 30 days), 70.8 percent were current illicit drug users and 62.9 percent were current marijuana users. By contrast, the respective percentages among youth who were not current alcohol users were 4.8 percent for current illicit drug use and 3.7 percent for current marijuana use (2023 NSDUH).

Although teen alcohol use has declined over the past three decades, the average annual number of deaths from excessive alcohol use among persons under 20 years of age continued to rise between 2016–2017 and 2020–2021 (Esser et al. 2024). Miech and colleagues (2025) found that among 12th graders surveyed in 2024, 8.8 percent reported consuming 5 or more alcoholic drinks, 3.0 percent reported consuming 10 or more drinks, and 1.7 percent reported consuming 15 or more drinks in a row at least once in the last 2 weeks. This type of extreme binge drinking, in particular, places youth at heightened risk for a range of serious outcomes, including injuries, alcohol-related suicide, homicide, sexual assault, alcohol poisoning, motor vehicle crashes, other drug use, alcohol use disorder, memory blackouts, altered brain development, poor academic performance (Evans-Polce et al. 2017; Hingson and White 2013, 2014; Hingson and Zha 2018; Kruckow et al. 2023; Patrick and Azar 2018; White and Hingson 2013; White et al. 2011), and adverse effects on new verbal learning (Carbia et al. 2017; Lees et al. 2020; Nguyen-Louie et al. 2016; Rodríguez Holguín et al. 2023).

Adolescents are particularly sensitive to the neurotoxic effects of alcohol because their brain development is not fully complete (Crews et al. 2007; de Goede et al. 2021; Kuhns et al. 2022; Spear 2018; Sun et al. 2023). Alcohol consumption in adolescence disrupts normal brain maturation in several regions (e.g., prefrontal cortex, hippocampus, and cerebellum), resulting in neurobehavioral and cognitive deficits (Courtney et al. 2019; Cservenka and Nagel 2016; Guerri and Pascual 2010; Jacobus and Tapert 2013; López-Caneda et al. 2019, 2024; Pascual et al. 2018; Ruan et al. 2019; Spear 2014, 2016, 2018; Squeglia et al. 2009, 2014a, 2014b; Sullivan et al. 2020; Tapert and Eberson-Shumate 2022; Welch et al. 2013; White and Swartzwelder 2005; Zhao et al. 2020). Neuroimaging, neurophysiological, and neuropsychological studies collectively demonstrate that pre-existing deviations from the typical developmental trajectories during adolescence in key frontal and limbic/subcortical brain regions, along with other regions such as the insula and cerebellum, and their associated functional networks act as risk markers predisposing youth to drinking or binge drinking (Baranger et al. 2023; López-Caneda et al. 2019, 2024; Miller et al. 2024; Rane et al. 2022; Yip et al. 2023). Binge drinking during this critical period of vulnerability further exacerbates disruptions in adolescent brain development through structural, functional, and cognitive aberrations, such as accelerated gray matter volume reductions in cortical and subcortical regions; attenuated growth in white matter structures associated with axonal myelination; and alterations in brain activation and connectivity for cognitive control (executive functioning) tasks involving working memory, inhibition, and decision making (Carbia et al. 2018; Cservenka and Brumback 2017; Feldstein Ewing et al. 2014; Infante et al. 2022; Jones et al. 2018; Lees et al. 2019, 2020; Müller-Oehring et al. 2018; Pérez-García et al. 2022b; Pfefferbaum et al. 2018; Silveira et al. 2020; Squeglia et al. 2017; Zhao et al. 2020). These neural changes contribute to deficits in attention, learning, and visuospatial skills and may involve altered reward processing (Cservenka and Brumback 2017; Jones et al. 2018; Macht et al. 2020b; Tetteh-Quarshie and Risher 2023). The effects of binge drinking on brain development also show sex-specific differences among adolescents, with females experiencing greater inhibition of overall brain growth and increased microglial activation in the hippocampus, and males experiencing reduced cortical thickness, volumetric decline in reward-related regions, and functional deficits in brain areas associated with social behavior (Nwachukwu et al. 2022; Pérez-García et al. 2022a; Piekarski et al. 2022; Towner et al. 2023). Results from preclinical studies have shown long-lasting adverse effects of adolescent alcohol exposures on brain structure and function, neurogenesis, and cognitive and neurobehavioral outcomes in adulthood, mediated by neuroimmune, neurotrophic, epigenetic, and other neurobiological alterations at the molecular, cellular, and physiological levels (Coleman et al. 2014; Crews et al. 2016, 2019, 2024; Fernandez et al. 2016; Gass et al. 2014; Gilpin et al. 2012; Hauser et al. 2023; Hiller-Sturmhöfel and Spear 2018; Kyzar et al. 2016; Logrip et al. 2013; Macht et al. 2020a; Nwachukwu et al. 2023; Pandey et al. 2022; Semenova 2012; Spear 2016; Taffe et al. 2010; Tetteh-Quarshie and Risher 2023; Vetreno and Crews 2015; Vetreno et al. 2016).

A large body of literature has found that early age at initiation of drinking (i.e., the first consumption of a full drink) or intoxication is associated with future drinking patterns and alcohol-related risk behaviors (Blomeyer et al. 2013; Buchmann et al. 2009; Caetano et al. 2014; Dawson et al. 2007, 2008; Deutsch et al. 2013; Faden 2006; Guttmannova et al. 2011; Hermos et al. 2008; Hingson and Zha 2009; Hingson et al. 2002, 2006; Holligan et al. 2019; Kim et al. 2017; Lee et al. 2012; Liang and Chikritzhs 2013; Maimaris and McCambridge 2014; McBride et al. 2014; Newton-Howes et al. 2019; Pautassi et al. 2020; Sjödin et al. 2024; Stueve and O’Donnell 2005; Swahn et al. 2008, 2013; Warner and White 2003; Wells et al. 2004; York et al. 2004; Young et al. 2006; Zakrajsek and Shope 2006). For example, early age at drinking initiation is associated with past-year drinking (Lee et al. 2012), frequency and quantity of drinking (Aiken et al. 2018; Deutsch et al. 2013), volume of drinking (Caetano et al. 2014), binge drinking (Aiken et al. 2018; Buchmann et al. 2009; Caetano et al. 2014; Liang and Chikritzhs 2013), extreme binge drinking (Livingston et al. 2023), an earlier onset of hazardous drinking (Gardner et al. 2024), getting drunk and high (Stueve and O’Donnell 2005), alcohol misuse (Kim et al. 2022), alcohol use disorders later in life (Caamano-Isorna et al. 2020; Caetano et al. 2014; Dawson et al. 2008; Grant et al. 2001; Guttmannova et al. 2011; York et al. 2004), substance use in earlier adulthood (Brennan et al. 2025), prescription drug misuse (Arterberry et al. 2016; Hermos et al. 2008), comorbid reports of physical fighting and suicide attempts (Swahn et al. 2013), and dating violence victimization and perpetration (Swahn et al. 2008). Moreover, early age at initiation of drinking is associated with driving after drinking and being involved in drinking-related vehicle crashes (Hingson et al. 2002), risky driving and alcohol-related driving offenses (Zakrajsek and Shope 2006), unintentionally injuring oneself and others when under the influence of alcohol (Hingson and Zha 2009), and other risky and delinquent behaviors (Livingston et al. 2023). It is also associated with increased risk for memory loss (Livingston et al. 2023), alcohol-related neurocognitive vulnerabilities, poor neuropsychological functioning (Nguyen-Louie et al. 2017), and alterations in frontoparietal functional connectivity (Nguyen-Louie et al. 2018).

There is substantial geographic variation in underage drinking across states, regions, and substate areas (Center for Behavioral Health Statistics and Quality 2025). Among youth ages 12 to 20, the state prevalence of past-month alcohol use in 2022–2023 ranges from 10.1% in Utah to 20.1% in Massachusetts; past-month binge drinking ranges from 5.7% in Utah to 12.2% in Wisconsin. Perceptions of great risk associated with having 5 or more drinks once or twice a week also vary, from 32.9% in Vermont to 45.5% in Maryland (2022–2023 NSDUH).

Data Sources

Data for this report are drawn from three sources: the NSDUH, the MTF survey, and the YRBS.

National Survey on Drug Use and Health

The NSDUH (formerly the NHSDA) is conducted by the Substance Abuse and Mental Health Services Administration (SAMHSA). The nationally representative survey was initiated in 1971 and has been administered at the household level annually since 1991. The data were historically collected through in-person interviews. However, the survey introduced multimode data collection procedures in October 2020, permitting respondents to complete the survey via the web or in person in eligible locations. Since 1999, computer-assisted interviewing methods have been used, including audio computer-assisted self-interviewing for selected survey components during in-person data collection. In 1999, the survey sample was expanded to allow for computation of state-specific prevalence estimates. In 2002, the survey was given its new name (NSDUH), and additional methodological changes were made that affected some prevalence rates and represented a new baseline. These included a $30 incentive to all respondents that resulted in substantial increases in response rates and improved data quality-control measures. In 2002, 2011, and 2021, new population data from the 2000, 2010, and 2020 decennial Censuses, respectively, became available for use in NSDUH sample weighting procedures. Unlike previous designs that divide the sample approximately equally among three age groups—12 to 17 years, 18 to 25 years, and 26 years or older—the 2014/2015 design places more sample (50%) in the 26 years or older age group to provide more accurate estimates for the aging drug use population. The partial questionnaire redesign in 2015 changes the threshold of binge alcohol use from 5 or more drinks for both sexes to 5 or more drinks for males and 4 or more drinks for females on an occasion in the past 30 days. Beginning in 2021, NSDUH fully transitioned from traditional in-person interviews to multimode data collection, and these methodological changes in survey administration create a discontinuity for trend comparisons with prior years. All respondents ages 12–20 years from the NSDUH public-use data were selected for this surveillance report.

Monitoring the Future

The MTF survey is funded by the National Institute on Drug Abuse and conducted annually by the University of Michigan Institute for Social Research (Miech et al. 2025). The MTF survey was initiated in 1975 among 12th graders only, and 8th and 10th graders were added in 1991, providing a nationally representative sample of secondary-school students in those grades. The survey is administered in school to a sample of students enrolled in public and private secondary schools, using multiple questionnaire forms, with four forms in 8th and 10th grade and six forms in 12th grade. Since 2019, the MTF in-school survey has introduced an electronic format, gradually replacing paper-and-pencil questionnaires. Starting in 2021, students complete a web-based questionnaire on their own Internet-connected electronic devices. Since 1976, a random sample of 12th graders has been followed biannually through a self-administered mail-back questionnaire, with web-based surveys introduced in 2018. Beginning in 2005, MTF survey introduced a measure of extreme binge (high-intensity) drinking, defined as consuming 10 or more or 15 or more drinks in a row within the past 2 weeks, in the 12th‑grader surveys. For this surveillance report, all respondents from the 8th, 10th, and 12th grade samples were included in the analyses.

Youth Risk Behavior Survey

The YRBS is conducted by the Centers for Disease Control and Prevention and administered every 2 years to students enrolled in public or private schools in grades 9 through 12. Initiated in 1991, the survey provides a nationally representative sample of youth enrolled in high school. Since 2023, the national YRBS has transitioned from paper-and-pencil to electronic survey administration. Beginning in 2017, YRBS changed the threshold of binge alcohol use from 5 or more drinks for both sexes to 5 or more drinks for males and 4 or more drinks for females in a row in the past 30 days. All respondents from the biannual surveys were included in this surveillance report.

All three sources of data for this surveillance report are nationally representative repeated cross-sectional surveys that provide data on youth alcohol consumption and related risk behaviors. As described above, there are also important differences among the three surveys related to the ages of youth sampled, the timing of the survey, the setting of survey administration and consequent type of youth sampled and level of anonymity involved, and the wording of the questions. Specifications of the three surveys are summarized in the Appendix and briefly reviewed here.

Age Groups

Whereas NSDUH collects information on youth as young as age 12, the YRBS collects data from students in grades 9 through 12, and MTF skips grades, surveying students in 8th, 10th, and 12th grades.

Periodicity

Both NSDUH and MTF are administered annually, whereas the YRBS collects data every 2 years.

Survey Administration Location

As school-based surveys, MTF and YRBS collect data only on youth currently enrolled in school. Data from NSDUH cover both youth enrolled in school and those not enrolled. The different settings of the survey administration contribute to different levels of anonymity. For example, youth may feel more comfortable revealing sensitive information, such as alcohol and other drug use, in the more anonymous school setting than at home (Faden et al. 2004; Fendrich and Johnson 2001; Fowler and Stringfellow 2001; Sudman 2001). Additionally, youth may respond to perceived peer influence in the school setting and thus may exaggerate certain risk behaviors in their self-reports. Such effects are hard to assess (Fowler and Stringfellow 2001; Harrison 2001). Differences in survey setting, administration methods, and respondent privacy likely contribute to lower estimates of adolescent alcohol consumption in the household-based NSDUH (due to underreporting) than in the school-based MTF and YRBS, where potential overreporting may occur (Gnambs and Kaspar 2015).

Question Wording

The three surveys differ in the number and wording of questions on alcohol consumption and on related risk behaviors. The reference period for all surveys is generally the past 30 days. However, MTF asks about binge drinking in the past 2 weeks, whereas the other surveys ask about it in the past month. NSDUH and YRBS ask about age at initiation of drinking, but MTF asks about grade at initiation of drinking. All surveys ask about driving under the influence, but MTF asks only the 12th graders. MTF frames the question around whether the respondent has ever received a moving violation ticket or warning after drinking alcohol. The Appendix specifies the wording used in each survey for the indicators included in the report. Data tables and figures are not provided for all questions listed.

Methods

This surveillance report tracks alcohol consumption and associated behaviors among youth ages 12–20 years old. Age 12 is the youngest age for which nationally representative data are available (from NSDUH). Age 20 is the last year before youth are legally allowed to drink alcohol. Findings are presented for 1991 through 2023, the latest year for which data are available for this report from all three data sources. However, for some indicators introduced in later years, only data available since their inception are presented (e.g., detailed data by race, which has been collected since the 1999 NSDUH).

Definitions

The report presents trend data on different categories of indicators for alcohol consumption and related risk behaviors, including prevalence of use, pattern of use, alcohol-related attitudes, and alcohol-related risk behaviors. Selected indicators from the most recent NSDUH data (Center for Behavioral Health Statistics and Quality 2025) are also presented by state in Figures 10 through 12. Definitions of measures used, including descriptions of calculated measures, are provided below:

• Prevalence of use
  • Any drinking of alcohol in the past 30 days (more than just a sip or two from a drink)
• Initiation of drinking
  • Age at first use of alcohol
  • Initiating drinking at age 12 years or younger
• Frequency of use
  • Number of drinking days in the past 30 days
• Quantity of use
  • Usual number of drinks on drinking days in the past 30 days
• Volume of use
  • Total number of drinks in the past 30 days, computed as frequency multiplied by quantity in the past 30 days
• Binge drinking
  • Any drinking of 5 or more drinks on an occasion (in a row) in the past 30 days (for NSDUH 1991–2014 and YRBS 1991–2015) or in the past 2 weeks (for MTF); any drinking of 5 or more drinks for males and 4 or more drinks for females on an occasion in the past 30 days (for NSDUH 2015–2023 and YRBS 2017–2023)
  • Frequency of drinking 5 or more drinks (for NSDUH 1991–2014) or 5 or more drinks for males and 4 or more drinks for females (for NSDUH 2015–2023) on an occasion in the past 30 days
  • Frequency of drinking 5 or more drinks (for NSDUH 1991–2014) or 5 or more drinks for males and 4 or more drinks for females (for NSDUH 2015–2023) on an occasion in the past 30 days, categorized as 0 days, 1–2 days, 3–4 days, and 5 or more days
• Extreme binge (high-intensity) drinking (i.e., twice or three times as much as the traditional 5-drink cutoff)
  • Any drinking of 10 or more drinks in a row in the past 2 weeks
  • Any drinking of 15 or more drinks in a row in the past 2 weeks
• Drunkenness
  • Having been drunk or very high from drinking in the past 30 days
• Alcohol-related attitudes
  • Disapproval of consuming 1 or 2 drinks nearly every day, categorized as don’t disapprove, disapprove, and strongly disapprove
  • Disapproval of consuming 5 or more drinks once or twice each weekend, categorized as don’t disapprove, disapprove, and strongly disapprove
  • Perception of risk of harm for those drinking 1 or 2 drinks nearly every day, categorized as no risk, slight risk, moderate risk, and great risk
  • Perception of risk of harm for those drinking 5 or more drinks once or twice each weekend, categorized as no risk, slight risk, moderate risk, and great risk
• Alcohol-related risk behaviors
  • Driving after drinking alcohol in the past 30 days
  • Riding in a car, within the past 30 days, driven by someone who had been drinking
  • Drinking alcohol or using drugs before last sexual intercourse

Analyses

Analyses in this report are mainly descriptive. The report examines a wide range of relevant data over multiple years and discusses observed changes without reference to more complicated statistical analyses of trends, such as those used by Faden and Fay (2004) and Faden (2006). The exceptions are Figures 1-1, 1-2, 6-1, and 6-2, where the trends in prevalence of drinking and binge drinking are represented by segmented lines that were fitted using joinpoint (piecewise) regression models (Kim et al. 2000). These lines indicate significant changes in the trends over time.

For different analyses, data are presented by sex, race (non-Hispanic white and non-Hispanic black) and Hispanic origin, age or current school grade, and school enrollment status. More detailed data by race and Hispanic origin (non-Hispanic white, non-Hispanic black, Native American/Alaska Native, Asian/Native Hawaiian Islander/Other Pacific Islander, Hispanic, more than one race) are presented in the tables for 1999–2023. To increase the stability of estimates, prevalence estimates by age are presented in age groups, with ages grouped into categories of 12–14, 15–17, and 18–20. Three-year moving averages are used for tracking certain alcohol consumption measures to minimize data suppression problems for groups with small sample sizes (e.g., number of drinkers in the 12–14 age group) and to avoid large fluctuations in point estimates. For selected indicators, trend data from the three data sources are compared in separate graphs. Because of the large age span of the NSDUH and its inclusion of youth enrolled and not enrolled in school, more detailed analyses by demographic characteristics are presented for the NSDUH only. We present results from secondary data analysis for NSDUH and YRBS data, because publicly available reports on these surveys do not cover all the indicators, categories, and age groupings applicable to this report. When applicable estimates by grade are publicly available from MTF reports, we present them in tables to facilitate comparison across different surveys. We present our secondary analysis results based on public-use MTF data only when the aggregate estimates are not publicly available elsewhere. Because some MTF questions are asked only of subsamples, we note in the MTF figures and tables where estimates are based on data from subsamples.

To ensure the reliability of the results presented, extreme values were adjusted using a winsorization approach. Specifically, unusually high values were capped at the 95th percentile when calculating the average number of drinks consumed on drinking days (i.e., quantity) and across all days in the past year (i.e., volume). Cases with missing values for an indicator of interest were excluded from analysis for that indicator.

To enable readers to assess the precision of the estimates provided, each estimate is accompanied by a value for the standard error of the estimate (labeled S.E. in the tables). Multiplying the standard error by 1.96 provides a margin of error above and below each estimate. This range defines a 95-percent confidence interval that implies that 95 percent of such calculated intervals over repeated independent sampling are expected to contain the true value being estimated. Estimates with very large standard errors can be extremely unreliable. The reliability of NSDUH estimates was evaluated according to the NSDUH suppression criteria, which consider prevalence, relative standard error (defined as the ratio of the standard error to the estimate), nominal sample size, and effective sample size for each estimate (Center for Behavioral Health Statistics and Quality 2024). Estimates deemed to have low reliability are denoted in the tables with the symbol #.

Limitations

Due to differences in sample populations (i.e., only school-enrolled youth vs. both enrolled and not enrolled youth; 12- to 20-year-olds for NSDUH vs. 8th, 10th, and 12th graders for MTF vs. 9th–12th graders for YRBS) and survey administration (i.e., in school vs. at home), comparisons among the three data sources should be made with caution. Numerous factors contribute to lower estimates of adolescent alcohol use in NSDUH than in MTF or YRBS and lower estimates in MTF than in YRBS, including interview privacy, survey focus, prominence of mentions of substance use, procedures for obtaining parental permission, assurances of anonymity or confidentiality, placement and context of substance use questions in the interview, survey mode, and structure and wording of survey questions (SAMHSA 2012). Despite differences in prevalence estimates across surveys within a single year, it is possible to compare trends across years. For example, there are marked differences in absolute values of past-30-day alcohol consumption prevalence among findings from NSDUH, MTF, and YRBS. However, trend lines from all three surveys show a decrease in prevalence between 1997 and 2023.

Furthermore, NSDUH had major design changes during this period. Close examination of the NSDUH data for past-30-day consumption shows a steep decrease in rates between 1998 and 1999 followed by an increase from 1999 to 2003. It is not clear whether this is a real increase or an artifact of the 1999 methodological changes. The 2002 methodological changes, including the respondent incentive and the improved data collection quality-control measures, also may have resulted in higher self-reported substance use by respondents (SAMHSA 2003). Caution should be taken when comparing estimates across survey years, particularly before and after 1999, 2002, and 2021, due to methodological changes that may affect trend comparability. The major change in survey methodology of 1999 and the additional changes of 2002 and 2021 are marked with shading in the graphs in this report. Beginning in 2015, the partial questionnaire redesign lowered the threshold of binge drinking for females from 5 or more drinks to 4 or more drinks in a row in the past 30 days and could potentially affect the trends by artificially inflating the prevalence estimates for females and for both sexes combined. This also applies to YRBS beginning in 2017. Prevalence estimates from NSDUH included in this report may differ slightly from those presented in reports issued by SAMHSA, as SAMHSA analysts use a restricted-use dataset for their analyses. The same is true with estimates generated from our secondary analysis of public-use MTF data.

Due to COVID-19 pandemic-related delays and adjustments, the 2021 YRBS was administered primarily during the fall semester instead of its traditional spring semester schedule. NSDUH paused data collection in March 2020, resumed web-based interviews in October 2020, and fully implemented multimode data collection including both web-based and in-person interviews beginning in 2021. The 2020 MTF survey was not fully conducted because of school closures and pandemic disruptions, which severely limited the scope and reliability of estimates for that year. Recent methodological changes in survey administration and data collection, such as the gradual transition to electronic modes across all three surveys starting in 2019, along with COVID-19-related disruptions and associated biases, may affect the comparability and reliability of estimates from these surveys. Nonetheless, the overall convergence in findings from these surveys consistently establishes robust downward trends in underage drinking.

List of Tables

See the pdf for full tables.

Table 1-1. NSDUH: Prevalence of drinking in the past 30 days, by age, sex, and race/Hispanic origin, among 12- to 20-year-olds, United States, 1991–2023

Table 1-2. MTF: Prevalence of drinking in the past 30 days, by grade and sex, United States, 1991–2023

Table 1-3. YRBS: Prevalence of drinking in the past 30 days, by grade and sex, United States, 1991–2023

Table 2-1. NSDUH: Mean age at first use of alcohol, by age, sex, and race/Hispanic origin, among ever drinkers ages 12–20, United States, 1991–2023 (based on 3-year moving averages)

Table 2-2. NSDUH: Prevalence of initiating drinking at age 12 or younger, by sex, among ever drinkers ages 12–20, United States, 1991–2023

Table 2-3. YRBS: Prevalence of initiating drinking at age 12 or younger, by sex, among ever drinkers, United States, 1991–2023

Table 3. NSDUH: Mean frequency of drinking in the past 30 days, by age, sex, and race/ Hispanic origin, among current drinkers ages 12–20, United States, 1991–2023 (based on 3-year moving averages)

Table 4. NSDUH: Mean quantity of drinking on drinking days in the past 30 days, by age, sex, and race/Hispanic origin, among current drinkers ages 12–20, United States, 1991–2023 (based on 3-year moving averages)

Table 5. NSDUH: Average total number of drinks in the past 30 days, by age, sex, and race/Hispanic origin, among current drinkers ages 12–20, United States, 1991–2023 (based on 3-year moving averages)

Table 6-1. NSDUH: Prevalence of binge drinking in the past 30 days, by age, sex, and race/Hispanic origin, among 12- to 20-year-olds, United States, 1991–2023

Table 6-2. MTF: Prevalence of binge drinking in the past 2 weeks, by grade and sex, United States, 1991–2023

Table 6-3. YRBS: Prevalence of binge drinking in the past 30 days, by grade and sex, United States, 1991–2023

Table 6-4. NSDUH: Mean number of binge drinking days in the past 30 days, by age, sex, and race/Hispanic origin, among current drinkers ages 12–20, United States, 1991–2023 (based on 3-year moving averages)

Table 7. MTF: Prevalence of having been drunk or very high from drinking alcoholic beverages in the past 30 days, by grade and sex, United States, 1991–2023

Table 8. NSDUH: Prevalence of driving a vehicle while under the influence of alcohol or in combination of illegal drugs in the past 12 months, by age, sex, and race/Hispanic origin, among 12- to 20-year-olds, United States, 1995–2023 (based on 3-year moving averages)

Figures

Figure 1-1. Prevalence of drinking in the past 30 days, by sex, 1991–2023

Figure 1-2. NSDUH: Prevalence of drinking in the past 30 days among 12- to 20-year-olds, by sex and age group, 1991–2023

[Fitted trends result from joinpoint regression (see the methods section for detail).]

[Shaded areas mark the implementation of a major redesign to the NSDUH survey in 1999 and methodological changes in 2002 and 2021.]

Figure 1-3. NSDUH: Prevalence of drinking in the past 30 days among 12- to 20-year-olds, by sex and race/Hispanic origin, 1991–2023

Figure 1-4. NSDUH: Prevalence of drinking in the past 30 days among 12- to 20-year-olds, by school enrollment status, 1991–2023

[Shaded areas mark the implementation of a major redesign to the NSDUH survey in 1999 and methodological changes in 2002 and 2021.]

Figure 1-5. NSDUH: Prevalence of drinking in the past 30 days among 12- to 20-year-olds, by age, sex, and race/Hispanic origin, 2023

Figure 1-6. YRBS: Prevalence of drinking in the past 30 days, by grade and sex, 2023

Figure 2-1. NSDUH and YRBS: Prevalence of initiating drinking at age 12 or younger among ever drinkers, by sex, 1991–2023

Figure 2-2. NSDUH: Mean age at first use of alcohol among 12- to 20-year-olds, by sex and race/Hispanic origin, 1991–2023 (based on 3-year moving averages)

[Shaded areas mark the implementation of a major redesign to the NSDUH survey in 1999 and methodological changes in 2002 and 2021.]

Figure 2-3. NSDUH: Mean age at first use of alcohol among 12- to 20-year-olds, by school enrollment status, 1991–2023

Figure 3-1. NSDUH: Mean frequency of drinking in the past 30 days among current drinkers ages 12–20, by sex and race/Hispanic origin, 1991–2023 (based on 3-year moving averages)

[Shaded areas mark the implementation of a major redesign to the NSDUH survey in 1999 and methodological changes in 2002 and 2021.]

Figure 3-2. NSDUH: Mean frequency of drinking in the past 30 days among current drinkers ages 12–20, by age, sex, and race/Hispanic origin, 2019–2023 (based on 5-year averages)

Figure 4-1. NSDUH: Mean quantity on drinking days in the past 30 days among current drinkers ages 12–20, by sex and race/Hispanic origin, 1991–2023 (based on 3-year moving averages)

[Shaded areas mark the implementation of a major redesign to the NSDUH survey in 1999 and methodological changes in 2002 and 2021.]

Figure 4-2. NSDUH: Mean quantity on drinking days in the past 30 days among current drinkers ages 12–20, by age, sex, and race/Hispanic origin, 2019–2023 (based on 5-year averages)

Figure 5-1. NSDUH: Average total number of drinks in the past 30 days among current drinkers ages 12–20, by sex and race/Hispanic origin, 1991–2023 (based on 3-year moving averages)

[Shaded areas mark the implementation of a major redesign to the NSDUH survey in 1999 and methodological changes in 2002 and 2021.]

Figure 5-2. NSDUH: Average total number of drinks in the past 30 days among current drinkers ages 12–20, by age, sex, and race/Hispanic origin, 2019–2023 (based on 5-year averages)

Figure 6-1. Prevalence of binge drinking, by sex, 1991–2023

[Fitted trends result from joinpoint regression (see the methods section for detail).]

[The shaded area marks the implementation of a major redesign to the NSDUH survey in 1999 and methodological changes in 2002 and 2021.]

Note: The threshold for females was changed to consumption of 4 or more drinks on an occasion in the past 30 days beginning in NSDUH 2015 and YRBS 2017.

Figure 6-2. NSDUH: Prevalence of binge drinking in the past 30 days among 12- to 20-year-olds, by sex and age group, 1991–2023

[Fitted trends result from joinpoint regression (see the methods section for detail).]

[Shaded areas mark the implementation of a major redesign to the NSDUH survey in 1999 and methodological changes in 2002 and 2021.]

Figure 6-3. NSDUH: Prevalence of binge drinking in the past 30 days among 12- to 20-year-olds, by age, sex, and race/Hispanic origin, 2023

Note: The threshold for females was changed to consumption of 4 or more drinks on an occasion in the past 30 days beginning in NSDUH 2015.

Figure 6-4. YRBS: Prevalence of binge drinking in the past 30 days, by grade and sex, 2023

Figure 6-5. NSDUH: Mean number of binge drinking days in the past 30 days among current drinkers ages 12–20, by sex and race/Hispanic origin, 1991–2023 (based on 3-year moving averages)

[Shaded areas mark the implementation of a major redesign to the NSDUH survey in 1999 and methodological changes in 2002 and 2021.]

Note: The threshold for females was changed to consumption of 4 or more drinks on an occasion in the past 30 days beginning in NSDUH 2015 and YRBS 2017.

Figure 6-6. NSDUH: Mean number of binge drinking days in the past 30 days among current drinkers ages 12–20, by school enrollment status, 1991–2023 (based on 3-year moving averages)

[The shaded area marks the implementation of a major redesign to the NSDUH survey in 1999 and methodological changes in 2002 and 2021.]

Note: The threshold for females was changed to consumption of 4 or more drinks on an occasion in the past 30 days beginning in NSDUH 2015.

Figure 6-7. NSDUH: Frequency of binge drinking in the past 30 days among current drinkers ages 12–20, by category (0 days, 1–2 days, 3–4 days, 5+ days), sex, and race/Hispanic origin, 1991–2023 (based on 3-year moving averages)

[Shaded areas mark the implementation of a major redesign to the NSDUH survey in 1999 and methodological changes in 2002 and 2021.]

Note: The threshold for females was changed to consumption of 4 or more drinks on an occasion in the past 30 days beginning in NSDUH 2015.

Figure 6-8. MTF: Prevalence of binge drinking (5+ drinks in a row) and extreme binge (high-intensity) drinking (10+ and 15+ drinks in a row) in the past 2 weeks among 12th graders, 2005–2023

Note: The question of 5+ drinks in a row was asked of full sample, but questions of 10+ and 15+ drinks in a row were asked of one-sixth of the sample.

Figure 7. MTF: Prevalence of having been drunk or very high from drinking alcoholic beverages in the past 30 days among 8th, 10th, and 12th graders, by sex, 1991–2023

Note: The question was asked of only about half the 8th and 10th grade sample for 1991–1996. It was asked of the full sample for 1997–2023.

For 12th graders, the question was asked of about one-third of the sample for 1991–2023.

Figure 8-1a. MTF: Percent distribution of disapproval toward taking 1 or 2 drinks nearly every day among 8th, 10th, and 12th graders, by sex, 1991–2023

Note: The question was asked of only about half the 8th and 10th grade sample for 1991–1996. It was asked of the full sample for 1997–2023.

For 12th graders, the question was asked of about two-thirds of the sample for 1991–2023.

Figure 8-1b. MTF: Percent distribution of disapproval toward having 5 or more drinks once or twice each weekend among 8th, 10th, and 12th graders, by sex, 1991–2023

Note: The question was asked of only about half the 8th and 10th grade sample for 1991–1996. It was asked of the full sample for 1997–2023.

For 12th graders, the question was asked of about two-thirds of the sample for 1991–2023.

Figure 8-2a. MTF: Percent distribution of perceived risk of harm by having 1 or 2 drinks nearly every day among 8th, 10th, and 12th graders, by sex, 1991–2023

Note: The question was asked of only about half the 8th and 10th grade sample for 1991–1996. It was asked of the full sample for 1997–2023.

For 12th graders, the question was asked of about five-sixths of the sample for 1991–2023.

Figure 8-2b. MTF: Percent distribution of perceived risk of harm by having 5 or more drinks once or twice each weekend among 8th, 10th, and 12th graders, by sex, 1991–2023

Note: The question was asked of only about half the 8th and 10th grade sample for 1991–1996. It was asked of the full sample for 1997–2023.

For 12th graders, the question was asked of about five-sixths of the sample for 1991–2023.

Figure 8-2c. NSDUH: Percent distribution of perceived risk of harm by having 5 or more drinks once or twice a week among 12- to 20-year-olds, by sex, 1991–2023

[Shaded areas mark the implementation of a major redesign to the NSDUH survey in 1999 and methodological changes in 2002 and 2021.]

Figure 9-1a. NSDUH: Prevalence of driving a vehicle while under the influence of alcohol or in combination of illegal drugs in the past 12 months among 12- to 20-year-olds, by sex and race/Hispanic origin, 1995–2023

[Shaded areas mark the implementation of a major redesign to the NSDUH survey in 1999 and methodological changes in 2002 and 2021.]

Figure 9-1b. YRBS: Prevalence of driving after drinking alcohol in the past 30 days,
by sex and race/Hispanic origin, 1991–2023

Figure 9-1c. YRBS: Prevalence of riding in a car driven by someone who had been drinking in the past 30 days, by sex and race/Hispanic origin, 1991–2023

Figure 9-2. YRBS: Prevalence of drinking alcohol or using drugs before last sexual intercourse, by sex, 1991–2023

Figure 10. NSDUH: Prevalence of drinking in the past 30 days among 12- to 20-year-olds, by state, 2022–2023

Figure 11. NSDUH: Prevalence of binge drinking in the past 30 days among 12- to 20-year-olds, by state, 2022–2023

Figure 12. NSDUH: Prevalence of perceptions of great risk from having five or more drinks of an alcoholic beverage once or twice a week among 12- to 20-year-olds, by state, 2022–2023

Appendix

Appendix A. Differences Among Survey Data Sources

See PDF for full appendix.