The present study investigated how intervention might alter the relationship between perpetrating violence and later drug use. A cluster-randomized controlled trial design involving 72 schools (38 intervention, 34 control) and 6390 students attending grades 7 and 8 was employed in Brazil. Drug use and violence were assessed at three points. A random-intercept cross-lagged panel model examined the reciprocal association between drug use and school violence domains across the three data collection waves. For both groups, we found that the cross-lagged effect of perpetration on further drug use in adolescents was stronger than the reverse, but the interrelationship was not statistically significant between #Tamojunto and control schools. The carry-over effects of drug use and violence were also not significantly different between groups. There is a lack of evidence showing that #Tamojunto can modify the dynamics between drug use and school violence across the 21-month period. The direction of the causal effect (i.e., the more perpetration behavior, the more subsequent drug use behavior) is present, but weak in both groups. The trial registration protocol at the national Brazilian Register of Clinical Trials (REBEC) is #RBR-4mnv5g.
