Patients were randomly allocated to treatment with either low energy pulses, high energy pulses, or high energy sinusoidal wave ECT and, despite the induction of convulsions of similar duration, were found to respond at different rates and to a different degree, depending on whether low or high energy currents were used. It is suggested that the quantity of current, as well as the induction of a convulsion, is relevant to therapeutic outcome with ECT. A reexamination of previous work shows that this is not necessarily inconsistent with these findings, which also explain the contradictory results in recent controlled trials of ECT.