Three alternative monitoring methods for assessing the duration of seizures during electroconvulsive therapy were studied. The mean integrated amplitude of the electroencephalogram, facial muscle electromyogram, and "cuff method" were compared with the reference single-channel unprocessed electroencephalogram in 78 sessions with 17 patients. The measures of seizure duration differed significantly (p < 0.001). The mean integrated electroence phalographic amplitude differences were small and an artifact of the sampling procedure. Larger, but clinically unimportant, discrepancies were obtained with the facial electromyogram amplitude. In contrast, there were marked differences between the electroencephalogram and the "cuff method," which suggest that the latter technique may be of limited usefulness.