BACKGROUND Clinical data have shown that evaluation of sinus rhythm variability is a useful test for selection of high-risk patients after myocardial infarction. The purpose of this study was to test the hypothesis that measures of ventricular rate variability (VRV) during non-sustained ventricular tachycardia (NSVT) may be helpful in predicting sudden death in patients with a remote myocardial infarction. METHODS The study group comprised 162 patients with remote myocardial infarction who showed NSVT on 24 h ECG recording. Temporal changes in up to 10 beat-to-beat V-V intervals of NSVT episodes were assessed. The following parameters of VRV-NSVT were calculated: average value of successive differences in V-V intervals (ADVV); normalized average value of successive differences in V-V intervals (nADVV). RESULTS During a mean follow-up of 4 years, 57 patients (35%) had died; sudden death was identified in 34 (21%). Patients who died suddenly had significantly (P < 0.001) lower values of ADVV and nADVV variables compared with survivors or those whose death was not sudden. The incidence of sudden death was assessed in the group of patients with more stable cycle durations in the episodes of NSVT (expected to be at high risk) and in the group with more variable cycle durations (expected to be at low risk). The relative risk of sudden death for patients with ADVV values less than 30 ms was 3.2 compared with those with ADVV values of at least 30 ms (P < 0.001), and that for patients with nADVV values less than 6% was 3.0 compared with those with nADVV values of at least 6% (P < 0.001). CONCLUSIONS These results indicate a relationship between regularity of NSVT rhythm and risk of sudden death. The assessment of VRV-NSVT can be used to identify patients with NSVT who remain at high risk after myocardial infarction.