A 72% reduction in the number of premature ventricular complexes, a 75 and 65% reduction in pairs and repetitive forms, respectively, are commonly accepted criteria for the evaluation of the efficacy of short-term antiarrhythmic treatment. On the other hand, a three-to-ten fold increase in the frequency of premature ventricular complexes is considered as a proarrhythmic effect. The aim of this paper is to verify if these criteria can still be applied on a long-term basis. Twenty-eight subjects without a demonstrable underlying organic heart disease and 21 patients with stable chronic ischaemic heart disease were studied. All patients showed more than 30 premature ventricular complexes per hour during ambulatory electrocardiogram monitoring. The spontaneous long term variability of ventricular arrhythmias was evaluated by comparing the first of three consecutive 24-hour ambulatory ECGs at the beginning of the study with a 24-hour ECG monitoring at the end of a 24-month follow-up period. According to the previous criteria about 50% of patients showed a spontaneous reduction in the number of premature ventricular complexes (greater than 72%) mimicking a therapeutic effect if any drug had been given. The increase in ventricular arrhythmias mimicked a proarrhythmic effect in almost 10% of patients. This trend is substantially the same both in subjects without organic heart disease and in patients with chronic coronary artery disease. The results of this study demonstrate that the initially established criteria used to judge the efficacy or the proarrhythmic effect of a given drug cannot be relied upon on a long-term basis.(ABSTRACT TRUNCATED AT 250 WORDS)