The ability of exercise testing to predict the extent of coronary artery disease was examined in 268 male patients undergoing both coronary angiography and bicycle testing with electrocardiography before coronary artery bypass surgery. When maximal ST-depressions limited by symptoms increased from 0 to 4 mm or more, the percentage of patients with 'serious' coronary disease, defined as either triple vessel disease or left main stem stenosis, increased from 50% to 80% (P = 0.0001). The patients in the lowest third of physical work capacity showed only a slightly increased risk of serious disease. This tendency was abolished in patients who were using beta-blockers, whereas the relationship between ST-depression and disease was not affected by this medication. The probability of finding left main stem stenosis in a patient increased from 5 to 30% with increasing ST-depression: beta-blockers did not affect this relationship, but there was no additional predictive effect of implicating the level of physical work capacity. It is concluded that traditional electrocardiography during exercise is of value when selecting patients for angiography, but that the physical work level obtained during the test does not predict the degree of coronary pathology.