The study began in 1976 with a group of 1,634 students in the 7th, 8th, and 9th grades. A latent-variable structural modeling method was employed to capture the general tendency to use drugs while retaining unique aspects of cocaine, alcohol, and other illicit drug use. Prevalence rates across the 8-year study revealed that alcohol remained consistently high, illicit drug use increased during adolescence and decreased into young adulthood, and cocaine use increased steadily through adolescence into young adulthood. No sex differences in prevalence were noted. Despite the general conclusion that cocaine involvement and general drug use were not highly distinguishable, at least during adolescence, several small but intriguing differential effects were observed. Only the number of times super-high on cocaine was directly predicted from emotional distress variables; times super-high on cocaine at year five was directly predicted from low cheerfulness in year one. The frequency of cocaine use in young adulthood was uniquely predicted by an 8-year earlier antecedent of illicit drug use, as well as by late adolescent alcohol use. Thus, involvement with illicit drugs at a very early age was a precursor for much later cocaine involvement. Early psychological distress increased intense cocaine use and contributed to drug problems. 61 references, 4 tables, and 2 figures.