OBJECTIVE: To discuss how to develop more effective treatment programmes than those currently available for addictive disorders. DATA SOURCES: The Swedish SBU report, published in English in 2003, was used as a database. It includes 641 randomized controlled trials and seven longitudinal prospective studies. METHODS: Meta-analytical calculations were performed in several areas using standardized mean differences (d) effect-size estimate and homogeneity testing. Three critical issues have been the focus of the present analysis: the early intervention phase, treatment procedures and their additive properties and the transitional period between early and late effects of treatment. RESULTS: The main findings while integrating the results in a new way were that intervention studies with one single session showed a small but robust homogeneous effect size, whereas studies of interventions with several sessions were heterogeneous with large and small effect sizes among the included studies. Similar effect sizes were found in alcohol, opioid and cocaine treatment studies. Agonist treatment yielded the highest effect sizes. Some evidence was found for a possible additive effect for cognitive behaviour therapy and naltrexone as well as for aversive treatment (disulfiram) and psychosocial treatment in alcohol dependence. So far studies on the transition period between short- and long-term outcome are few and inconclusive. CONCLUSIONS: There is a prospect of improving addiction treatment, and the following areas are suggested by meta-analysis for future research: (a) to examine in more detail the process between the first and second session of intervention; (b) to randomize simultaneously for independent categories of psychosocial and psychopharmacological treatment; and (c) to intensify studies on the transitional period between short- and long-term outcome.