To assess QT interval dispersion on the surface electrocardiogram in patients with sustained ventricular arrhythmias. A retrospective and prospective blinded controlled study of patients referred for investigation of ventricular arrhythmias at a tertiary cardiac centre. 89 consecutive patients with sustained ventricular arrhythmias due to chronic ischaemic heart disease, cardiomyopathy, or ventricular tachycardia (VT) in a normal heart. 32 patients did not meet the inclusion criteria; therefore 57 patients were compared with a control group of 40 patients with myocardial disease but no history of arrhythmias and 12 normal controls with no myocardial disease. Standard 12 lead electrocardiograms were enlarged, the QT intervals for each lead measured, and QT dispersion calculated. There was a significantly greater mean QT dispersion (77 ms) in patients with sustained ventricular arrhythmias compared with the control group (38 ms, p < 0.01). This held for all groups; after myocardial infarction VT (82 (22) ms v control 38 (10) ms; p < 0.01), dilated cardiomyopathy VT (76 (18) ms v control 40 (11) ms, p < 0.01), and normal heart VT (65 (7) ms v control 32 (8), p < 0.05). There was also a greater QT dispersion in patients with impaired left ventricular function and VT, with a correlation between left ventricular function and QT dispersion in patients with VT (r = 0.56, p < 0.01). QT interval dispersion may be a further non-invasive marker of susceptibility to ventricular arrhythmias.