Abstract 70 patients with endogenous depression, defined by strict criteria, who fulfilled the Newcastle indications for electroconvulsive therapy (ECT) were randomly allocated either to a course of eight simulated ECTs or to a course of eight real ECTs. The improvement in terms of psychiatrists' ratings in the group of patients given real ECT was significantly greater (p < 0·01) than that in those given simulated ECT, but the difference between the two groups was small in relation to the considerable improvement of both groups over the 4-week treatment period. No differences were found between the two groups at one-month and six-month follow-up. The therapeutic benefits of electrically induced convulsions in depression were of lesser magnitude and were more transient than has sometimes been claimed. In the real-ECT group memory was impaired during treatment but memory tests revealed no difference between the groups at six-month follow-up.