In a group of 52 patients electroconvulsive treatment by means of a Thymatron apparatus was provided under simultaneous EEG control. In addition to parameters of the EEG curve (type of curve according to intensity of the paroxysmal process, duration of electroconvulsion on EEG, type of termination of paroxysm of EEG) also other variables were investigated in relation to the therapeutic effect. The following were identified as significant predictors for marked clinical improvement: type of EEG curve with more marked paroxysmal activity, shorter cumulative duration of electroconvulsions on EEG before first clinical improvement of the patient and a higher CGI entering score. Other variables proved unimportant, e.g. the patient's age, total number of electroconvulsions, total cumulative duration of electroconvulsions on EEG and type of termination of electroconvulsions on EEG.