AIMS: To assess the effectiveness of interventions directed at the prevention or reduction of use of illicit substances by young people or those directed at reducing harm caused by continuing use. DESIGN: A systematic review was conducted. Reports were identified through electronic and hand searching and contact with known workers in the area. Studies were included if they reported evaluations of interventions targeting illicit drug use and provided sufficient detail of the intervention and design of the evaluation to allow judgements to be made of their methodological soundness. Meta-analyses were conducted combining the data of the methodologically sound studies. PARTICIPANTS AND SETTINGS TARGETED BY INTERVENTIONS: Evaluations of interventions were included if their targeted audience included young people aged between 8 and 25 years. Identified evaluations were delivered in a range of settings including: schools and colleges; community settings; the family; medical/therapeutic settings; mass media. MEASUREMENTS: Data extracted from each report included details of design, content and theoretical orientation of intervention, setting of the intervention, target audience, methods, population size, subject refusal rates, rates of attrition, outcome measures, length of follow-up and findings, including statistical power. FINDINGS: The majority of studies identified were evaluations of interventions introduced in schools and targeting alcohol, tobacco and marijuana simultaneously. These studies were methodologically stronger than interventions targeting other drugs and implemented outside schools. Meta-analyses showed that the impact of evaluated interventions was small with dissipation of programme gains over time. Interventions targeting hard to reach groups have not been evaluated adequately. CONCLUSIONS: Effort needs to be directed towards the development of improved evaluative solutions to the problems posed by these groups. There is still insufficient evidence to assess the effectiveness of the range of approaches to drugs education; more methodologically sound evaluations are required. There is also a need to target interventions to reflect the specific needs and experiences of recipients.