AIMS: To investigate the short- and long-term effect of psychological treatments of pathological gambling and factors relating to treatment outcome. DESIGN AND SETTING: This study provides a quantitative meta-analytical review of psychotherapeutic treatments of pathological gambling. Studies were identified by computer search in the PsycINFO and Medline databases covering the period from 1966 to 2004, as well as from relevant reference lists. INCLUSION CRITERIA: The target problem was pathological gambling, the treatment was psychological, the study was published in English and outcomes directly pertaining to gambling were employed. Single case studies, studies where elimination of gambling not was the priority and studies with insufficient statistical information were excluded from the present meta-analysis. PARTICIPANTS: A total of 37 outcome studies, published or reported between 1968 and 2004, were identified. Of these 15 were excluded, thus 22 studies were included, involving 1434 subjects. The grand mean age was 40.1 years. The overall proportion of men was 71.5%. MEASUREMENTS: The included studies were coded for outcome measures of pathological gambling. For each condition, means and standard deviations for gambling-related outcome measures, all based upon self-reports or therapist ratings, were compiled at three points in time: baseline, post-treatment and the last follow-up reported. FINDINGS: Effect sizes represent the difference between the mean score in a treatment condition and a control condition or the difference between mean scores at separated points in time for one group, expressed in terms of standard deviation units. At post-treatment the analysis indicated that psychological treatments were more effective than no treatment, yielding an overall effect size of 2.01 (P < 0.01). At follow-up (averaging 17.0 months) the corresponding effect size was 1.59 (P < 0.01). A multiple regression analysis showed that the magnitude of effect sizes at post-treatment were lower in studies including patients with a formal diagnosis of pathological gambling only, compared to studies not employing such inclusion criteria. Effect sizes were also higher in randomized controlled trials compared to not randomized controlled trials, higher in within subjects designs compared to between subjects designs and also positively related to number of therapy sessions. No mediator variables were significantly related to the magnitude of the effect sizes at follow-up. CONCLUSION: Psychological interventions for pathological gamble seem to be yield very favourable short- and long-term outcomes.