The accuracy of coronary cineangiography in predicting the degree of stenosis in coronary arteries was evaluated by comparing autopsy and premortem cineangiographic findings in 25 patients. Coronary cineangiograms and autopsy specimens were reviewed independently by two cardiologists and two pathologists. Identical diagrams dividing the cononary arteries into 12 segments were used by both groups to record the location and degree of stenosis observed. Cineangiographic findings were in agreement with pathologic findings (less than 25 percent difference in cross-sectional luminal area) in 178 (79 percent) of the 226 segments examined, but overestimated the degree of stenosis in 13 (6 percent) and underestimated it in 34 (15 percent). Thus, cineangiography appears to be a reliable tool in evaluating coronary artery disease. When diagnostic errors are made, they are usually underestimations of the degree of disease; common causes of error are circumferential stenosis, eccentric lesions, obstruction of view by artifical valves and poor opacification due to severe proximal stenosis.