In vivo studies were performed on 28 dogs to evaluate the usefulness of transmission computed tomography (CT) in the detection and quantitation of experimentally induced myocardial infarction. Intravenously administered contrast material was required to define the internal structure of the heart and to differentiate normal from infarcted tissue. Transmural infarcts with homogeneous central regions were visualized as areas of diminished contrast enhancement compared with the normal myocardium. All transmural infarcts of at least 24 hours' duration showed a surrounding border zone of patchy necrosis that was variable in size and had high CT numbers due to slow washout of the contrast material from this region. Infarct area determined from the images for individual slices correlated well (r = 0.976) with that calculated using pathology. The technique is very sensitive and can detect infarction within a papillary muscle. Nontransmural or patchy infarcts show up as areas of diffuse contrast enhancement without a central core of diminished enhancement. The distribution of the contrast material is similar to that of technetium-99m pyrophosphate in the border zone of the infarct in infusion studies, but in bolus studies it behaves more like thallium-201.