Three methods of electroconvulsive therapy (ECT) were compared in respect of therapeutic effect in 69 attacks of endogenous depression in 65 patients, not previously treated by ECT during the actual period of illness. The treatments were given under barbiturate narcosis, with full muscular relaxation, administration of oxygen and electroencephalographic recording of the seizure discharge. In two methods grand mal seizures were evoked by supraliminal (A) and liminal stimulation (B), in the third (C) lidocaine (3 mg/kg i.v.) was given before the application of liminal stimulation. The seizure discharges in C were markedly shortened and their pattern modified, while between A and B the duration and pattern of the seizures were similar (Table 1). The patients were referred to the three treatments at random and the groups may be regarded as having a similar prognosis (Table 2-5, 7). The therapeutic outcome was estimated by rating several depressive symptoms according to a rating scale worked out for the purpose. The rating was performed before treatment, one week after the fourth treatment (a treatment pause was then made) and one week after the completed series. To obtain more reliable measures the scores for the various symptoms were added together to form a total score, which was then divided into a depression score and a retardation score, presumably measuring mainly depressive mood and psychomotor retardation. Differences in rating scores on two rating occasions were taken as measures of improvement. In addition, a global rating of improvement was made. The rating procedure was double blind. The principal results were: 1. After four treatments (three patients who recovered after three treatments included) the degree of improvement was in the rank order ABC with significant group differences for a few scores. After the completed series of treatments improvement in groups A and B did not differ significantly whereas in group C it was significantly smaller for some scores (Table 9). 2. The total number of treatments was significantly higher in group C than in group A and B, which did not differ significantly between themselves (Table 10). 3. A measure of therapeutic efficiency, improvement per treatment, was computed by dividing the degree of improvement as obtained from the differences in the combined scores and from the global score of improvement, by the number of treatments. After four treatments the improvement per treatment was highest in group A and lowest in group C, although in the comparisons AB and BC most differences did not reach significance. After the completed series the improvement per treatment did not differ significantly in groups A and B whereas in group C it was significantly less (Table 10). 4. In comparison with groups A and B, the total duration of seizure discharges was significantly shorter in group C both after four treatments and, in spite of the higher number of treatments, after the completed series. The improvement per second of seizure discharge was not significantly different in the groups although there was a tendency to a lower effect per second in group B (Table 11). It is concluded from these results that shortening of the seizure discharge decreases the therapeutic efficiency of ECT. Increase of the stimulus intensity, which apparently does not change the seizure discharge, possibly gives a more rapid therapeutic response but does not change the final degree of improvement or the number of treatments required to reach it. The depression-relieving effect of ECT is bound to seizure activity and not, or only slightly, to other effects of electrical stimulation.