A computer and a hand search of the literature recovered 33 papers from which 25 trials suitable for meta-analysis were identified. We compared the effectiveness of cognitive-behavioural treatments with the waiting list control and alternative treatment control conditions. There was a great diversity of measurements which we grouped into domains representing major facets of pain. Effect sizes, corrected for measurement unreliability, were estimated for each domain. When compared with the waiting list control conditions cognitive-behavioural treatments were associated with significant effect sizes on all domains of measurement (median effect size across domains = 0.5). Comparison with alternative active treatments revealed that cognitive-behavioural treatments produced significantly greater changes for the domains of pain experience, cognitive coping and appraisal (positive coping measures), and reduced behavioural expression of pain. Differences on the following domains were not significant; mood/affect (depression and other, non-depression, measures), cognitive coping and appraisal (negative, e.g. catastrophization), and social role functioning. We conclude that active psychological treatments based on the principle of cognitive behavioural therapy are effective. We discuss the results with reference to the complexity and quality of the trials.