The influence of the underlying heart disease on the spontaneous variability of ventricular arrhythmias was investigated prospectively in 53 patients (25 CHD, 28 IDC) with frequent and complex ventricular arrhythmias. In each patient, two consecutive ambulatory 24-h Holter ECGs were prepared and in each tape the mean hourly arrhythmia count (AC) was determined separately for singular VPCs, couplets, and salvos. The spontaneous variability between the two long-term ECGs was defined as the logarithm of the ratio (ACday 2 + 0.01)/(ACday 1 + 0.01). The 95% confidence intervals of the stated types of arrhythmias were calculated as +/- 2 SD. The results were analyzed as a function of the underlying etiology, NYHA class, and left ventricular ejection fraction. There were no differences between patients with CHD and IDC. The extent of left ventricular dysfunction did not have any influence either. In patients of NYHA class 3 there was a higher spontaneous variability of VPCs, couplets and salvos than in patients of NYHA class 2, but the differences could not be ensured statistically. We conclude from the results that the validation of an antiarrhythmic treatment can be performed independently from the nature of the underlying heart disease and the left ventricular ejection fraction. However, it remains unclear whether a greater variability must be expected in patients of NYHA class 3 than in patients of NYHA class 2.