The prognosis of patients with heart disease and prediction of sudden cardiac death can be assessed through heart rate variability, an indirect measure of abnormal autonomic control. The authors have evaluated the heart rate variability by 24-hour ambulatory electrocardiographic monitoring in 25 diabetic patients, 19 ischemic heart disease patients, 18 congestive heart failure patients, and 10 normal subjects. Thirteen diabetic patients had autonomic neuropathy and 12 patients did not. Heart rate variability index (mean SD) in patients with diabetes mellitus, ischemic heart disease, and congestive heart failure was significantly lower (34.5 +/- 12.6 ms, 43.7 +/- 15.4 ms, and 34.6 +/- 15.8 ms vs 65.6 +/- 16.7 ms, p < 0.05) than that of normal subjects. Mean SD was significantly lower in patients with autonomic neuropathy as compared to patients without autonomic neuropathy (26.4 +/- 6.5 ms vs 44.2 +/- 11.0 ms, p < 0.05). Patients who died suddenly had a significantly lower (p < 0.05) mean SD as compared to survivors: 49 +/- 7 ms in patients with mild ischemic heart disease, 48 +/- 15 ms in patients with severe ischemic heart disease, and 23 +/- 7 ms in patients who died. Similarly, the mean SD in 4 congestive heart failure patients who died was lower significantly (p < 0.05) than in those who survived (19.0 +/- 5.6 ms vs 40.0 +/- 14.5 ms). Among congestive heart failure patients, clinical improvement by therapy was associated with a significant increase in mean SD. When the mean SD of 30 ms was used as the cutoff point for detection of autonomic dysfunction or patient death, specificity exceeded 90% and sensitivity was 75%. The mean SD of less than 30 ms reflects either an impaired cardiac autonomic nerve innervation or an imbalance in the autonomic nerve activity, or both. This value of mean SD might be a good marker in clinical practice concerning autonomic dysfunction and poor prognosis in patients with diabetes mellitus, ischemic heart disease, or congestive heart failure.