Aims: Although crack cocaine first appeared in cities in the United States in the mid-1980s, little is known about its use over long periods of time. This study identified crack cocaine user groups on the basis of long-term trajectories.

Design: Following a natural history approach, data were collected periodically from 1996 to 2005. Group-based modeling assessed the probability of a crack smoker becoming abstinent during the observation period.

Setting: A targeted sampling plan guided the recruitment of a community sample of crack cocaine users in Dayton, Ohio.

Participants: Crack smokers (n = 430) 18 years or older whose urine tested positive for cocaine metabolites at the baseline interview.

Measurements: Interviewer-administered and audio computer self-administered, structured questionnaires were used to collect data on a range of variables, including frequency of crack use. Abstinence was defined as not having used crack for at least 6 consecutive months during the study.

Findings: Three trajectory-based groups were identified: (1) No Change, characterized by a very low probability of abstinence; (2) Some Change, characterized by a low to moderate probability of abstinence; and (3) Dramatic Change, characterized by a high probability of abstinence. African Americans and men were significantly less likely to become abstinent. For the majority of the people (63.6%), crack use was uninterrupted by extended periods of abstinence during the study.

Conclusion: Crack cocaine use that persists for a decade or longer may well be the norm for a large proportion of people who have experience with the drug.