Gateway Drug Theory is the hypothesis that using one intoxicating substance leads to future use of another. The theory is intuitively plausible, because there is certainly a strong one-way correlation between certain pairs of drugs. For example, 72% of people who have ever used cocaine tried marijuana first, whereas only 1% of marijuana users tried cocaine first. However, correlation is not causation; it is impossible to determine how many of the cocaine users might never have tried it had they declined that first joint.

Still, the correlation data is very striking; if we cannot use it to define gateway drugs, we can analyze it to determine typical pathways between substances of abuse. The U.S. federal Substance Abuse and Mental Health Services Administration (SAMHSA) performs an extremely detailed study of American drug use every year, the National Survey on Drug Use and Health (NSDUH), and one section records the age at which respondents first tried different intoxicants. With this data, we can visualize the exact steps that people took from one substance to another.

First, we’ll look at people the survey flagged as a current “abuser of or dependent on” each of nine intoxicants. The survey tracks whether respondents have ever tried, even once, any of these nine drugs, plus seven more. The first row (gray) shows people who have a problem with a substance yet have never tried any other substance of abuse in their lives; as you might imagine, this is pretty uncommon. Only people with alcohol problems have a significant number who’ve never tried anything else on the list, and even that’s only 13%.

By contrast, if you look at the second, dark blue, row, 78% or more of abusers of every drug studied have tried alcohol at least once. Given alcohol’s availability and social acceptability, that’s not surprising.

Looking down a column shows us which substances have ever been used by an abuser of that intoxicant. For example, if you look down the heroin column, you’ll see that they tend to have experimented quite a bit with other drugs, especially alcohol, marijuana, and painkillers.


There are too many exact pathways between drugs to track them all; for example, some respondents used alcohol, then marijuana, then cocaine and stopped there, while others had more complex and convoluted paths. But it is possible to visualize what were the substances tried immediately before and after every other substance. For example, 88% of people who have ever tried alcohol had never tried another intoxicant before; that’s true for only 1% of cocaine users. While 50% of people who have ever drunk alcohol have gone on to try another intoxicant, 34% have consumed marijuana at some point following their alcohol consumption. (If you click on marijuana in the left or right column, you can see how many of them have gone on to try other substances.)

Pay special attention to the “no further drugs tried” category, which varies a lot, and be careful not to read too much into it: These are snapshots of people of different ages, and some of them may go on to try other drugs in the future. Also, keep in mind that all respondents had to be able to respond to a survey, so devastating reasons (like death) for a drug being the last one tried are not counted.



The 2012 NSDUH survey was analyzed according to the ages at which people first tried substances, using SAMHSA’s weighting factors to extrapolate respondents to the U.S. population. Ties were divided up evenly; for example, if a respondent reported trying alcohol and marijuana for the first time during the same age, his or her weighting factor was evenly divided between alcohol, then marijuana, and marijuana, then alcohol. All results were rounded to the nearest percentage point, and results under 0.5% were discarded.

Embed These Interactive Graphics

Previous substance use among drug and alcohol abusers:
(minimum width: 630px)

Substances tried sequentially before and after:
(minimum width: 670px)

Fair Use encourages the use of the assets and results on this page. When including these results in any way, please link your audience to as well as this page so they may better learn about the methodology and see a full description of the data and its source.

Home | About T4A | Treatment Blog | Advertising | Sponsor Policy | Privacy Policy | Terms of Use | Site Map