Cannabis use peaked during late adolescence; liquor, stimulants, and nonprescription medication use increased steadily throughout the study period; hypnotics and psychedelic drug use increased through late adolescence and remained stable through young adulthood; and experimental use of cigarettes decreased to more committed use over the 8 years. In a latent variable model that did not include nonprescription drugs and cigarettes, alcohol use significantly influenced later cannabis and hard drug use, whereas cannabis use predicted later hard drug use for the earlier life period. Only the cannabis effect remained in the older age period, when drug use patterns appeared to become more stable and firmly established. However, when cigarettes and nonprescription drugs were included in the model, quite different results emerged. Results suggest that there are several minisequences of drug involvement. Cigarette use plays a prominent role as a gateway to cannabis and hard drugs, and at higher levels of involvement, there is a synergistic or reciprocal effect of increasing involvement. The later developmental period was much more stable than the earlier one and showed somewhat different patterns of influence. 3 figures, 9 tables, and 22 references. (Author abstract modified)