In a double-blind prospective study, 29 endogenously depressed patients (RDC) were randomized into sinusoidal wave (SW) and brief-pulse (BP) electroconvulsive therapy (ECT) groups. Bilateral modified treatments were administered on alternate days, three per week, and the treatment variables of current dosage and seizure duration were monitored for each treatment. Significantly more patients responded to SW than to BP ECT, but a comparable number of treatments was required to produce this response in the two groups. There was no difference in clinical or treatment variables between the SW and the BP groups, nor between ECT responders and nonresponders, with the exception that the SW-treated patients received larger doses of current per treatment than did the BP patients. For endogenous depression treated with ECT, we conclude that cumulative seizure duration may not be a parameter of significance, that overall rate of recovery in ECT responders is independent of stimulus waveform, and that some depressives may respond to SW but not to BP ECT. We suggest that the antidepressant effect of the ECT seizure may be characterized by a therapeutic window in current requirements; alternatively, a putative response threshold (again in terms of current requirement) may exist, which is higher in some patients than in others.