This study examined the effects of the Positive Action (PA) programme in Chicago Public Schools on problem behaviours among a cohort of elementary school students from grade three through grade five. Using a matched-pair, randomised control design with 14 elementary schools, approximately 510 fifth-graders self-reported lifetime substance use, serious violence-related behaviour, and current bullying and disruptive behaviours. Three-level (i.e. students nested within schools within school pairs) overdispersed Poisson models were used to examine programme effects on the number of items endorsed for each of the four outcomes. Findings indicated that students in the intervention endorsed 31% fewer substance use behaviours (incidence rate ratio [IRR] = 0.69), 37% fewer violence-related behaviours (IRR = 0.63) and 41% fewer bullying behaviours (IRR = 0.59), respectively, compared to students in the control schools. Reduction in reported disruptive behaviours was of a similar magnitude (27%, IRR = 0.73), but was not statistically significant. These results replicate findings of an earlier randomised trial of the PA programme and extend evidence of its effectiveness to youth attending large urban school systems.