Aims: There has been little available information on the long-term natural course, persistence and remission of cannabis use, abuse and dependence. The current study estimated rates and risk factors associated with stability and variation in cannabis use patterns, cannabis abuse and cannabis dependence in a community sample over a 10-year period. Design, setting and participants: Prospective longitudinal, epidemiological study with a 4- and 10-year follow-up of a community sample (n = 3021) aged 14-24 years at baseline in Munich, Germany. Measurements: Cannabis use, abuse and dependence and associated risk factors were assessed by face-to-face interviews using the Munich Composite International Diagnostic Interview. Findings: At baseline, one-third of the sample (34.2%) had used cannabis at least once. The cumulative incidence of cannabis use 10 years later was 50.7%. Fifty-six per cent of all repeated users (five times or more) at baseline reported cannabis use at 4-year follow-up. Ten years later, this proportion had decreased slightly to only 46.3%. Repeated (five times or more) users were almost three times more likely to report repeated use at 10-year follow up (OR = 2.8, 95% CI = 1.6-4.7), compared with those who had used cannabis fewer times. Peer use of cannabis, life-events and alcohol dependence also predicted use of cannabis at 10-year follow-up. Conclusions: Among youth who have used cannabis repeatedly (five times or more) cannabis use is fairly stable and rates of remission relatively low until age 34 yeras. Patterns of progression suggest that early targeted preventive measures should delay first use and reduce the number of experiences using cannabis, as these factors appear critical in progression to persistent cannabis use and cannabis dependence.