A system of 21-lead electrocardiography was used to assess 21 patients with severe angina during and after exercise using on-line computerised ST segment analysis. A direct comparison was made between the results obtained from 18 unipolar precordial leads and those from bipolar leads CM5 and CC5. Treadmill exercise was performed 48 hr prior to cardiac catheterization, which revealed luminal narrowing of at least 70% in one or more major coronary arteries in all cases. In all cases the ST depression exceeded 1 mm in both CM5 and CC5 at the peak of exercise. The magnitude of ST depression was greater in the bipolar leads in 75% of cases and in the remaining 25% the greatest peak ST depression occurred in a single unipolar lead. There was no correlation between the magnitude of ST depression and the number of coronary vessels involved. Isopotential surface mapping in the anterior, lateral and inferior projections from the unipolar leads at each stage of exercise failed to show a correlation between the area or distribution of ST segment change and the number or anatomical location of the vessels involved. It was not possible to show that the multiple-lead system could differentiate the site and severity of coronary artery disease in these patients with angina. The multiple-lead system was cumbersome and time-consuming in application and therefore cannot be recommended for routine exercise testing.