Background. Cognitive therapy (CT) has been studied in 78 controlled clinical trials from 1977 to 1996. Method. The meta-analysis used Hedges and Olkin d 1 and included 48 high-quality controlled trials. The 2765 patients presented non-psychotic and non-bipolar major depression, or dysthymia of mild to moderate severity. Results. At post-test CT appeared significantly better than waiting-list, antidepressants (P , 0.0001) and a group of miscellaneous therapies (P , 0.01). But, CT was equal to behaviour therapy. As between-trial homogeneity was not met, the comparisons of CT with waiting-list or placebo, and other therapies should be taken cautiously. In contrast, between-trial homogeneity was high for the comparisons of CT with behaviour therapy and antidepressants. A review of eight follow-up studies comparing CT with antidepressants suggested that CT may prevent relapses in the long-term, while relapse rate is high with antidepressants in naturalistic studies. Conclusion. CT is effective in patients with mild or moderate depression