The study used data from the RAND Adolescent Panel. The panel study was originally conducted to evaluate Project ALERT, a drug prevention program for middle school children (Ellickson and Bell, 1990). The baseline sample consisted of 6,527 seventh graders in 1985. Data for the current study were collected at grades 10 and 12 (or the equivalent, because the sample included school dropouts). Participants selected for the study had data available on drug use, emotional distress, and delinquent behavior at both grades 10 and 12. Structural equation modeling with latent variables showed strong cross-sectional correlations among drug use, delinquent behavior, and emotional distress; strong stability effects for all three problems; and only one cross-lagged effect (greater frequency of delinquent behavior at grade 10 led to greater problem drug use at grade 12). Multisample analyses by ethnicity (Asian, black, Latino, and white) showed that the cross-lagged effect was not statistically different in these four groups. The results suggest that curbing delinquent behavior might contribute to the prevention of problem drug use. 4 tables, 2 figures, and 54 references