To assess the accuracy of angiographic determinations of disease of coronary arteries and left ventricular myocardium we compared clinical with postmortem coronary arteriograms and left ventriculograms with myocardial pathology in 28 patients, all of whom died postoperatively and within three months of angiography; 19 had ischemic heart disease, four valvular heart disease, and five both. Comparison of pre and postmortem lumenal occlusion in 315 epicardial coronary segments, excluding those operated upon, showed greater than 50% narrowing discrepancies in 21 (7%). Significant coronary artery lesions were overestimated in six and underestimated in 15. Of the six overestimations, three appeared to be due to coronary spasm; of the 15 underestimations, 12 were due to overlapping images; six discrepancies were unexplained. Comparison of wall motion in 140 ventriculogram segments with myocardial pathology, excluding any post-study or perioperative injury, showed good correlation of reduced motion with 48 (34%) infarcted and 10 (7%) aneurysmal segments. However, 58 (41%) other segments had poor or absent ventriculogram motion, with structurally normal myocardium and patent coronary artery supply; 19 were on infarct margins and 39 in dilated or hypertrophied hearts. Thus, premortem coronary arteriographic occlusions generally indicate atherosclerotic narrowing; but decreased or absent segmental wall motion frequently does not indicate a myocardial lesion. It may be attributable to ischemia in the distribution of a critically narrowed coronary artery or it could be due to abnormal ventricular topography.