Myocardial infarcts are routinely detected by nuclear imaging techniques such as single photon emission computed tomography (SPECT) myocardial perfusion imaging. A newly developed technique for infarct detection based on contrast-enhanced cardiovascular magnetic resonance (CMR) has higher spatial resolution than SPECT. We postulated that this technique would detect infarcts missed by SPECT. We did contrast-enhanced CMR and SPECT examinations in 91 patients with suspected or known coronary artery disease. All CMR and SPECT images were scored, using a 14-segment model, for the presence, location, and spatial extent of infarction. To compare each imaging modality to a gold standard, we also acquired contrast-enhanced CMR and SPECT images in 12 dogs with, and three dogs without, myocardial infarction as defined by histochemical staining. In animals, contrast-enhanced CMR and SPECT detected all segments with nearly transmural infarction (>75% transmural extent of the left-ventricular wall). CMR also identified 100 of the 109 segments (92%) with subendocardial infarction (<50% transmural extent of the left-ventricular wall), whereas SPECT identified only 31 (28%). SPECT and CMR showed high specificity for the detection of infarction (97% and 98%, respectively). In patients, all segments with nearly transmural infarction, as defined by contrast-enhanced CMR, were detected by SPECT. However, of the 181 segments with subendocardial infarction, 85 (47%) were not detected by SPECT. On a per patient basis, six (13%) individuals with subendocardial infarcts visible by CMR had no evidence of infarction by SPECT. SPECT and CMR detect transmural myocardial infarcts at similar rates. However, CMR systematically detects subendocardial infarcts that are missed by SPECT.