Depressed patients were allocated randomly to individual cognitive therapy, group cognitive therapy or a waiting list `treatment as usual' control group. Blind clinical and psychometric assessment of patients revealed that those who underwent cognitive therapy did significantly better than those on the waiting list. There was no significant difference between patients treated with group or individual cognitive therapy. Threatment gains were maintained at follow-up at 12 months. Prognostic characteristics for the selection of depressed patients for cognitive therapy on the basis of the chronicity and social stresses are identified. It is concluded that cognitive therapy is an effective treatment which can be applied cost-effectively in general practice.