The accepted variation of 80-120 ms in the coupling interval of non-parasystolic ventricular premature beats has been linked to the duration of the preceding cycle length, but a lack of correlation between these factors has often been found. As a result, there is no explanation for variations in the coupling interval of such ventricular premature beats. To analyze the influence of the preceding cycle length on the coupling interval of isolated ventricular premature beats throughout an entire day, 10 otherwise healthy patients with frequent monofocal non-parasystolic isolated ventricular premature beats were studied with Holter recordings. A sample of the electrocardiogram was obtained at the beginning of each quarter of an hour and coupling intervals as well as preceding cycle lengths of ventricular premature beats not belonging to a bigeminy or trigeminy sequence were measured. Preceding cycle lengths changed in a way similar to the prevailing heart rate, with a marked prolongation during sleep. Coupling intervals showed an irregular pattern during the waking period but increased consistently from the first to the last hour of sleep. The average coupling interval during sleep was significantly longer than during waking (439 +/- 74 vs 466 +/- 80, p less than 0.01). In each patient a similar preceding cycle length within a narrow range of 40 ms was searched for during waking as well as during sleep and the corresponding coupling intervals were significantly longer during sleep (446 +/- 80 vs 466 +/- 85, p = 0.009).(ABSTRACT TRUNCATED AT 250 WORDS)