OBJECTIVE: To investigate the role of the autonomic nervous system in determining QT interval and dispersion. PATIENTS AND METHODS: 32 patients with chronic primary (idiopathic) autonomic failure (19 men, mean age 60 years) and 21 normal controls (11 men, mean age 59) without symptoms of ischaemic heart disease were studied retrospectively. Autonomic failure was diagnosed by a combination of symptomatic postural hypotension, subnormal plasma noradrenaline response to head-up tilt, and abnormal cardiovascular responses to standing, Valsalva manoeuvre, mental stress, cutaneous cold, isometric exercise, and deep breathing. QT intervals were measured from surface electrocardiograms and QT dispersion was defined as maximum QT--minimum QT occurring in any of the 12 leads. RESULTS: Mean heart rate (RR intervals) was similar in patients with autonomic failure and controls (S2 lead: 865 (132) v 857 (108) ms, P = NS; V2 lead: 865 (130) v 868 (113) ms, P = NS). QT intervals measured from electrocardiogram leads S2 and V2 were significantly longer in patients than in controls (401 (40) v 376 (16) ms, P < 0.01; and 403 (41) v 381 (20) ms, P < 0.05 respectively). The mean maximum QT interval in any lead, which is the best estimate of the maximum duration of electrical systole, was significantly longer in the patients than in controls (417 (48) v 388 (23) ms, P < 0.005). Linear regression analysis of QT and RR intervals for both groups showed a significant difference between the slopes of the two regression lines (F = 8.4, P < 0.001). However, QT dispersions were similar between patients and controls. CONCLUSIONS: Patients with primary autonomic failure have prolongation of QT intervals, indicating that the autonomic nervous system is an important determinant of QT interval. However, QT dispersion does not seem to be affected by chronic primary autonomic denervation.