The Brian C. Bennett Drug Charts

The Brian C. Bennett Drug Charts site comprises an extensive and easy-to-use collection of charts that present findings from decades of massive government survey data. This collection provides a more accurate and illuminating picture of drug use than is typically presented in popular media or reflected in our drug policies. Originally created by nonresident contributing expert Brian C. Bennett in 2002 and updated by the Baker Institute’s Drug Policy Program, these charts trace the pattern of the use and abuse of individual drugs over (in most cases) more than 40 years.

The most common of the charts show the percentage of people — either 12 and older or broken into smaller age groupings — who have ever used a given drug at some time in their lives, in the past year, and in the last month. High proportions of people who have ever used any of the drugs against which federal, state and local law enforcement agencies have waged war for more than 40 years stopped using them within the first year and no longer use them regularly, if at all.

We encourage visitors to dive in and start browsing through the charts.


  • Note the choice of age ranges, bottom left; choice of drugs, upper right. 
  • Choose Lifetime, Past Year, Past Month use by clicking on dots under the charts
  • Place the cursor at various points on any lines for use rates at a given time.
  • The little circles with two parallel lines at the ends of the gray bar at the top of the chart can be moved backward and forward to isolate periods of time. 

For a better experience, please read the brief introductory segments for helpful background and answers to some questions about the data. 


Introduction by William Martin

In the early 1970s, a National Commission on Marihuana and Drug Abuse, appointed by President Richard Nixon, urged federal and other levels of government to "maintain and monitor an ongoing collection of data necessary for present and prospective policy planning, including data on incidence, nature and consequences of drug use." That recommendation has been well satisfied. The National Survey on Drug Use and Health (NSDUH)[1], launched in 1971 and conducted by the Substance Abuse and Mental Health Services Administration (SAMHSA), administers face-to-face interviews with approximately 70,000 people ages 12 years and older annually. Monitoring the Future (MTF)[2], which launched in 1975 under the aegis of the National Institute on Drug Abuse (NIDA), questions approximately 50,000 teens annually, with follow-ups in later years. Both surveys ask a wide range of questions regarding drug use by interviewees, sorted by age, gender, ethnicity, geographic and urban-rural location, co-occurrence with mental health issues, types of drugs (licit and illicit) and frequency of use. These surveys, which easily qualify as Big Data, provide a rich and revealing picture of drug use in America, and their findings are consistently confirmed by other surveys. Still, although the information is available to anyone with a computer, few people appear to know the size or actual contours of drug use in this country, and the data play only a small role in public policy, mass media presentations and popular perception.

Until one becomes familiar with the format of the NSDUH, MTF and other valuable collections of statistics relevant to drug policy, quickly absorbing their implications can be daunting. Even after one gets the hang of it, the longitudinal graphs often cover only a modest number of years, making it difficult to trace trends across the nearly five decades of the War on Drugs. Beginning in 2002,  Brian C. Bennett, a former career intelligence analyst now at the University of Virginia, compiled and painstakingly sorted through a vast amount of official government data produced since 1970 regarding substance use and abuse. He then then displayed these data in easily understood graphs, charts, and tables that force one to recognize that many widely accepted beliefs and stereotypes about drug use in the United States are simply not true.

The most common of the charts show the percentage of people — either 12 and older or broken into smaller age groupings — who have ever used a given drug at some time in their lives, in the past year and in the last month. High proportions of people who have ever used any of the drugs against which federal, state and local law enforcement agencies have waged war for more than 40 years stopped using them within the first year and no longer use it regularly, if at all. When I have shown these charts to students, members of service clubs where I have spoken, or colleagues and supporters of the Baker Institute, the reaction is invariably surprise, sometimes approaching amazement. On occasion, some have told me that those charts did more than anything else to cause them to look at drug use and abuse and the War on Drugs in a quite different light.

To my initial disappointment, career demands caused Bennett to stop updating the charts after the 2008 surveys. Fortunately, when I contacted him, he said he would be pleased to have the Baker Institute Drug Policy Program share and update the work he had done and give it a home where the work could continue. In the ensuing period, Katharine Neill Harris, the Alfred C. Glassell, III, Fellow in Drug Policy, and  several interns updated a sizable number of the charts with data from the latest surveys, and the Institute’s graphic designers devised a new format for the charts, which Bennett approved. More recently, Farès El-Dahdah and Uilvim Ettore of the Humanities Research Center at Rice have used their remarkable skills to produce these more flexible and interactive charts.

At present, the charts and graphs on our website are a modest but growing fraction of those available on, which has received more than four million visits. We invite you to explore that site for more charts and many other fascinating features, with the warning that you may find it addictive. We expect to transfer more updated materials to our website in the near future, but always with explicit recognition of and appreciation for the enormous effort Brian Bennett, now a Contributing Expert to the Drug Policy Program, put into creating and maintaining the original site for many years.

[1]See 2019 SAMHSA NSDUH Report: This 71-page main document is easy to navigate due to a new format rolled out with the 2014 report. Much more extensive and detailed results can be found in the appendices. Until 2002, the name of the survey was National Household Survey on Drug Abuse. The name change reflects the recognition that not all use is abuse as well as greatly increased attention to mental health. 
[2] See


Introduction by Brian C. Bennett

I am pleased to join the Drug Policy Program at the Baker Institute and to welcome you to explore The Brian C. Bennett Drug Charts, which I began constructing in 2002. What you will find here is an extensive and easy-to-use collection of official government data, presented in a way that provides better context for creating a more accurate and complete picture of drug use and its impacts on our society. I have used only the best available and most accurate and complete data from U.S. governmental sources. The claims being made in the drug war debate can be properly evaluated only if the underlying data can be easily found and examined.

My intent has been to provide an encyclopedic presentation and analysis of the drug war that is easy for the average person to understand. Much that we are told about drugs and drug users is based on fear, grossly distorted and typically presented without any truly useful context. Our nation has been engaged in the War on Drugs for 100 years, but it is a battle against biology that can never actually be "won." It is a sign of intelligence to learn from experience. To that end, rather than producing another 100 years of the same thing, we clearly need to reconsider our approach to this issue. Indeed, even many of those once charged with waging the war are saying that it needs to stop. The situation is far worse than you may realize. Don’t simply believe what you read and hear. Look at the data yourself and tell other people about what you learn.


FAQs about the Survey Data

The best known of the Bennett charts draw on NSDUH and MTF data to trace and depict the use of various illegal drugs over decades by three variables: 1) ever used in one's lifetime, 2) used in the past year and 3) used in the past month. Looking at these charts clearly shows that the overwhelming majority of people who have ever used any illicit drug no longer do so regularly. An even smaller percentage have used the drug recently, indicating still lower levels of use that could reasonably signify dependence or addiction.

Any survey asking people to provide information about behavior that could possibly subject them to legal sanctions, social disapproval, or personal embarrassment or guilt has to answer at least three legitimate questions: 1) Who is being interviewed? 2) Who is not being interviewed? and 3) Are the people being interviewed telling the truth?

Who is being interviewed?

Most surveys cited in the media purport to reflect the opinions or other characteristics of a given population by following well-established protocols for selecting a random sample of its members to accurately represent the views or other characteristics of the whole population, within stated limits (e.g., “the margin of sampling error is ±3 percentage points at the 95 percent confidence level”). As a rule, the larger a well-drawn sample, the more accurate the results will be. The NSDUH sample of 70,000 people ages 12 and older and the MTF sample of 50,000 school-age children and young adults are enormous compared to the 700 to 1,500 people interviewed in most public opinion polls cited in the media. The NSDUH describes its population as follows:

In this report, terms such as “Americans,” “people in this country,” “general population” or similar terms are used broadly to refer to the civilian, noninstitutionalized population that is covered by NSDUH. Although some people in the general population of the United States are outside of the civilian, noninstitutionalized population, information from the 2010 census suggests that the civilian, noninstitutionalized population includes at least 97 percent of the total U.S. population.[1]

Who is not being interviewed?

The 3 percent of the population not sampled comprises homeless people; members of the military; people institutionalized in mental hospitals, facilities for the aged and otherwise infirm; and inmates in jails or prisons. Without question, many of these people have used or currently use drugs. But it would be a mistake to infer that including their usage would significantly alter the data drawn from the other 97 percent.

According to a 2015 U.S. Housing and Urban Development Department report, 564,708 people in the U.S. were considered homeless on a single night in January 2015, but 69 percent of those were in residential programs for homeless people and only 96,275 were regarded as “chronically homeless.” Nearly a fourth were children under the age of 18.[1] A significant number, as many as half in some reckonings, were on the streets for only a night or two.[2] It is not easy to gather solid data on the habits of the homeless, but a frequently cited SAMHSA study estimated that 38 percent of homeless people are dependent on alcohol and 26 percent abuse other drugs.[3]

If we accept a high estimate of the homeless population to be 500,000 and that 26 percent of these abuse drugs other than alcohol, the changes to the estimates in the NSDUH are minuscule. If all 130,000 homeless drug abusers use heroin regularly, clearly not the case, that would raise the percentage of past month users from 0.2 percent to 0.207 percent. If all 500,000 homeless people used heroin, the percentage would rise to only 0.34 percent. For cocaine and crack, the former less likely than heroin to be used by poor people because of its cost, the addition of a quarter of all homeless people would raise the current-user category by less than one-tenth of 1 percent.

If the sampled population had included the 1,354,054 members of the military, the result would likely be a lower percentage of current illicit drug use and a higher percentage of alcohol abuse, as a Google search for “military as alcohol culture” will attest. An Army report released in 2012 said that 43 percent of active duty soldiers had admitted to binge drinking in the previous month, almost twice the rate in the NSDUH.[4]

For obvious reasons, people institutionalized for reasons of age, mental illness or other infirmities are quite unlikely to use illicit drugs, but they may well use an unhealthy amount of prescription drugs. As for incarcerated people, in 2014 about half (95,800) of inmates in federal prison had been convicted of drug offenses,[5] while about 208,000 inmates in state prisons and more than 100,000 individuals in local jails were serving time for drug offenses. As with the numbers of homeless, if all these individuals were not incarcerated but were habitual users of any illicit drug, adding them to those who acknowledged their drug use to NSDUH interviewers would increase the total percent of current illicit drug users by less than 1 percent. It has been well established that African American and Hispanic users are far more likely to be incarcerated than whites even though their usage patterns differ little or not at all, so it is reasonable to think that their inclusion in the NSDUH population would have little effect on the reported results. But if we add all the 130,000 homeless people who use illicit drugs to all the 208,000 people incarcerated for using drugs, they would amount to just under 0.0012 of the U.S. population ages 12 and older.[6]

Nonresponders: In addition to the 3 percent of the population (ages 12 and older) not sampled, not every person asked to participate in the survey accepts the invitation. Some of those nonresponders live in high-crime neighborhoods, some are reluctant to talk about their drug use to a stranger, and, in a time when people are weary of responding to questions about how satisfied they are about their latest transaction with eBay or how they liked their recent visit with a physician they have been seeing for 20 years, some just don’t want to participate in any more surveys. These are issues worth noting, but survey research is a quite mature aspect of social science, and the enormous size of the NSDUH sample offers confidence that those who respond to the survey dependably represent the 97 percent of the population sampled.

[2] Dennis Culhane, “Five myths about America’s homeless,” Washington Post, July 22, 2010. Retrieved from:
[3] “Behavioral Health Services for People Who Are Homeless,” Substance Abuse and Mental Health Services Administration. Retrieved from:, p. 1.3.

Are the people being interviewed telling the truth?

A final understandable question remains: Are respondents telling the truth when they talk about their use of drugs, both legal and illicit, and other behaviors, such as mental health, that are measured by the NSDUH? Experts responsible for the survey and researchers dependent on the accuracy of its findings have for decades investigated such questions as the effects of age, race, gender, education, social standing, success in assuring confidentiality and skill of the interviewer.[1] Not surprisingly, the overall assessment is that although the methods of the NSDUH continue to improve, its findings are not and will never be perfect. The consistency of those findings over decades, however, offers considerable assurance that they provide a dependably accurate picture of drug use in America.

Whatever the flaws, NSDUH and MTF data are the best we have and the best we are likely to get. They are certainly, to use the common phrase, “close enough for government work.” They are routinely cited as authoritative by the White House Office of National Drug Control Policy (ONDCP), the National Institute on Drug Abuse (NIDA), the Centers for Disease Control and Prevention (CDC), the Partnership for Drug-Free Kids (formerly The Partnership for a Drug-Free America) and state and local health departments, as well as by newspaper, television and radio reporters and researchers at universities, NGOs and think tanks. These are the data of record, the numbers that must be used to formulate public policy. Unfortunately, for far too long, they have played only a small role in public policy, mass media presentations and popular perception. We hope, by widening exposure to these data, to alter that situation.

[1] For a sample of methodological studies conducted from 1971-2014, see


A Note on Survey Methodology and “Trend” Data

The Substance Abuse and Mental Health Services Administration’s (SAMHSA’s) Center for Behavioral Health Statistics and Quality oversees the development and implementation of the annual National Survey on Drug Use and Health (NSDUH). In order to obtain the most accurate data possible regarding mental health and substance use patterns in the U.S., SAMHSA periodically revises the NSDUH to reflect real-world changes such as emerging drug trends or new mental health policies.

Since its inception in 1971, the NSDUH has undergone several changes to meet demands for more and higher quality data. For example, in 1990 the questionnaire went from being administered every two or three years to being administered annually. In 1999, the pencil-and-paper survey was replaced with computer-based surveys. In 2002, interview respondents were offered $30 incentives to participate, and the name of the survey was changed from the National Household Survey on Drug Abuse to the National Survey on Drug Use and Health. Other revisions have included changing the way drugs are grouped together, adding new slang terms for various substances, and redefining features of certain behaviors, such as the number of drinks one has to consume to qualify as a binge drinker. (For more information about NSDUH survey redesigns and implications for trend data, click here.)

The majority of NSDUH survey changes are minor, but sometimes SAMHSA will declare them to have enough statistical significance that data collected with the new survey instrument are not comparable to data collected in previous years. This is referred to as “a break in trends.” Trend measurement is a valuable aspect of NSDUH data that allows comparison of behavior patterns over time. Maintaining trend data is an important goal of the NSDUH survey, one which must be balanced with the need to periodically revise the survey instrument to improve data quality.

SAMHSA often takes steps to preserve trend data following survey redesign. This typically involves using statistical methods to isolate the effects that changes in survey methodology may have on estimates for substance use and other behaviors, and to adjust estimates based on old survey designs to make them comparable to estimates based on the new survey instrument. (All of the data presented in the charts below reflect the most current estimates available.) But sometimes SAMHSA determines that a break in trend data is unavoidable. For example, SAMHSA maintains that NSDUH data beginning in 2002 is not comparable to years prior due to significant changes in methods, including the use of monetary incentives for a much larger sample than in years past.

SAMHSA also identifies changes to the 2015 survey as significant enough to warrant a break in trend data. The 2015 NSDUH questionnaire underwent revisions that affected several categories of substance use and mental health questions, including some of the drug categories for which data are presented on this website. Specifically, the hallucinogens drug category was broadened to include questions about use of ketamine, salvia and several other drugs that previously were asked about in a separate section; the slang term “Molly” was added to questions about Ecstasy; questions were added to the inhalants section regarding use of “air dusters,” slang for inhaling felt-tip pens and computer keyboard cleaner; questions about methamphetamine use, previously treated as prescription drug use in the survey, were changed to reflect that most methamphetamine use comes not from prescription methamphetamines (e.g., Desoxyn) but from methamphetamines that is produced on the black market; and several changes were made to the prescription drug categories to respond to growing concerns about prescription pill misuse, including the removal of questions about lifetime use, the addition of name-brand drugs such as Adderall and Ambien that previously were asked about elsewhere in the questionnaire, and the addition of a more detailed description of “drug misuse.”

Based on changes to these measures (discussed in greater detail here), SAMHSA cautions that 2015 data are not comparable to data for prior years for hallucinogens, Ecstasy, inhalants, methamphetamines and all prescription drug categories, including pain relievers, sedatives, stimulants and tranquilizers. (Measures for other drugs presented here, including marijuana, cocaine, crack, heroin, PCP, LSD and alcohol use, are not affected by the 2015 survey changes.) While we acknowledge SAMHSA’s concerns, the drug use data presented in the charts below include data for all years available, including 2015 and all other “trend break” years. We choose to maintain the trend data despite SAMHSA’s concerns for a few important reasons.

The effects of the NSDUH survey redesigns, while likely critical to statistical assessments of data, have less impact on practical or “common sense” interpretations of the data. For many of the drug categories presented here, the percentages of respondents who report drug use are so small relative to the overall population that any impact on these estimates from survey changes is, as Brian Bennett notes, “a case of splitting hairs.” Far more likely to affect estimates of use rates are issues of under-reporting drug use among survey participants, the lack of response from a portion of individuals asked to complete the survey, and the failure of the survey to capture populations at high risk for drug and alcohol use, such as prison and jail inmates and military personnel. These challenges to survey validity, discussed in greater detail in the above section, are independent of questionnaire revisions.

In short, the revisions to the 2015 NSDUH questionnaire and to questionnaires for prior years can have important implications for detailed statistical analysis of the data. But that is not our goal here. Rather, we present these data to highlight the “big picture” trends in drug use in the United States over the last 40 years. As we note elsewhere, these data, while imperfect, are the best we have, and are used as an authoritative source on drug use by federal, state and local government agencies and policymakers. To the extent that these data are used to inform public policy, they are done so irrespective of changes to survey redesign. Thus, it makes little sense, from a practical standpoint, for us not to present trend data for all years available. And survey changes do not at all impact two of the most critical takeaways from these data: that the War on Drugs is a battle waged against a problem that affects a small percentage of Americans, and that drug war policies have had little effect on drug use rates over the last five decades.

The Big Picture: Stable Rates, Alcohol and Age

Before looking at examples of the Bennett charts on individual drugs, it may be helpful to provide a graphic overview of drug use in America as reflected in the NSDUH. These data point clearly to three major aspects of substance abuse in the United States: 1) the stability of rates and patterns of drug use and drug problems over time, 2) the overwhelming role of alcohol and 3) the crucial importance of age.

Figure 1, drawn from NSDUH findings for 2002–2014, depicts the number of U.S. residents ages 12 or older estimated to have a diagnosable Substance Use Disorder (SUD),[1] an umbrella term for repeated patterns of harmful drug use, differentiating it from non-problematic use and indicating abuse and/or dependence, in the past 12 months. The substances monitored include alcohol, illicit drugs and nonmedical use of prescription drugs.

Figure 1: Substance Dependence or Abuse in the Past Year 12 years and older (2002-2014)

Note first the similarity in the height of the 13 columns.

Although tens of millions of people first used alcohol or illicit drugs during this period, the number of people with problematic use of alcohol only, of both alcohol and illicit drugs, or of illicit drugs only were remarkably similar each year, averaging slightly more than 22 million people annually despite an influx of new users. The number of people who developed problems with various substances was roughly matched by the number who recovered from them, usually on their own and without treatment.

Secondly, notice the predominance of alcohol abuse in each of these years.

On average over these 13 years, 68 percent of SUDs involved alcohol only, another 14 percent involved both alcohol and illicit drugs, and 18 percent involved illicit drugs only. It is important to recognize that alcohol is a drug with risk factors similar to other drugs. In fact, in 2010, the National Institute on Alcohol Abuse and Alcoholism charged alcohol with being the drug most likely to provoke violent behavior and cause the greatest loss of mental and physical control.

Figure 2 explains how this stability in usage rates can be true over time. Most people who use any potentially problematic drug don’t use it for long after initial experimentation, and even those who use it frequently, even to the point of abuse, are quite likely to reduce their use significantly as they age. Year after year, decade after decade, the pattern repeats itself. As seen in this chart, using data from the 2013 and 2014 NSDUH, illicit drug use begins as early as age 12 and accelerates rather sharply, reaching a peak between ages 18 and 20, which are the prime years for risky behavior, particularly by young men. It then drops rapidly over the next two decades, influenced by such factors as academic demands, family and career responsibilities, hangovers and awareness that addiction is not a desirable state.

Figure 2: Past Month Illicit Drug Use – 12 years or older (by age, 2013 and 2014)

The apparent major exception to this pattern is alcohol, by far the most widely used psychoactive drug in the U.S., encouraged by its ready availability and enormous sums spent to advertise its desirability. As Figure 3 depicts, more than 80 percent of people 12 and older in the United States have used alcohol at some point in their lives, two-thirds use it at least occasionally each year and more than half use it on a rather frequent basis.

Figure 3: Alcohol Use (1979-2014)

But despite high percentages of people who use alcohol in a non-problematic way throughout their lives, the patterns for binge drinking and heavy alcohol use follow an age-related decline similar to that of other psychoactive drugs, as seen in Figure 4.

Figure 4: Current, Binge and Heavy Alcohol Use – 12 years or older (by age, 2014)

Please keep these overall patterns of drug use and abuse in America in mind as you look at the individual charts in the Bennett collection, which we now invite you to do.

[1] For a detailed description of the criteria for diagnosing Substance Use Disorder, see “Rx for U.S. Drug Policy: A New Paradigm,” Baker Institute Policy Report Number 63, 2015, p. 12, note 7. Retrieved from

To understand how these data can inform smarter and more effective U.S. drug policy, please read our recent Issue Brief, which can be accessed here.

Usage Rates by Drug

Alcohol  /  Amphetamines  /  Cocaine  /  Crack Cocaine  /  Hallucinogens  /  Heroin  /   Inhalants  /  LSD  /   Marijuana  /   MDMA (Ecstasy)  /   Methamphetamines  /   Nonmedical Prescription Pills  /   Nonmedical Prescription Pain Relievers  /   PCP  /   Sedatives  /   Stimulants  /   Tranquilizers  

A PDF version of the Brian C. Bennett Drug Charts and the introductory materials found above can be accessed here. To gain access to the raw data, contact William Martin at The charts below were updated in June 2019 to reflect the most current available data.



Click any chart below to view as a larger image and to see how use rates differ by age group.

Back to the list of drugs


Click any chart below to view as a larger image and to see how use rates differ by age group.

Back to the list of drugs


Click any chart below to view as a larger image and to see how use rates differ by age group.

Back to the list of drugs

Crack Cocaine

Click any chart below to view as a larger image and to see how use rates differ by age group.

Back to the list of drugs


Click any chart below to view as a larger image and to see how use rates differ by age group.

Back to the list of drugs


Click any chart below to view as a larger image and to see how use rates differ by age group.

Back to the list of drugs


Click any chart below to view as a larger image and to see how use rates differ by age group.

Back to the list of drugs


Click any chart below to view as a larger image and to see how use rates differ by age group.

Back to the list of drugs


Click any chart below to view as a larger image and to see how use rates differ by age group.

Back to the list of drugs

MDMA (Ecstasy)

Click any chart below to view as a larger image and to see how use rates differ by age group.

Back to the list of drugs


Click any chart below to view as a larger image and to see how use rates differ by age group.

Back to the list of drugs

Nonmedical Prescription Pills

Click any chart below to view as a larger image and to see how use rates differ by age group.

Back to the list of drugs

Nonmedical Prescription Pain Relievers

Click any chart below to view as a larger image and to see how use rates differ by age group.

Back to the list of drugs


Click any chart below to view as a larger image and to see how use rates differ by age group.

Back to the list of drugs


Click any chart below to view as a larger image and to see how use rates differ by age group.

Back to the list of drugs


Click any chart below to view as a larger image and to see how use rates differ by age group.

Back to the list of drugs


Click any chart below to view as a larger image and to see how use rates differ by age group.

Back to the list of drugs