A group of patients with angina pectoris were investigated prospectively using a simple and well-recognised exercise test protocol, and S-T segment displacements during exercise were correlated with the results of coronary arteriography. Definitive exercise results in 100 patients correlated well with findings on coronary angiograms: 69 out of 70 patients with positive exercise test results had significant coronary disease on angiography, while 23 out of 30 with negative exercise test results had no significant coronary disease. In six patients a period of physical training changed indeterminate initial tests to definitive ones after the training. These results show that the use of a widely available exercise test and a light physical training programme, with rigid adherence to the protocol, can reduce the overall demand on coronary arteriography.