We investigated the efficacy of magnetic resonance imaging (MRI) in the detection and delineation of acute hemorrhagic cerebral infarction and evaluated the role of gradient-echo imaging in these patients. This study was performed prospectively. Entry criteria were: (1) Clinical evidence of acute supratentorial infarction later confirmed by at least one imaging study, and (2) unenhanced computed tomography (CT) and MRI scans performed within 72 hours of ictus. The first 50 patients who met these criteria were included. Comparison of CT and spin-echo (SE) and gradient-echo (GE) MRI at 0.6 T for the visualization of infarction and for the detection or exclusion of hemorrhage were done. There were no cases in which CT was superior to MRI for the diagnosis of either bland or hemorrhagic infarction. All 50 infarcts were detected on long TR SE MRI and 30/50 on GE scans. Eighteen infarcts were judged to be hemorrhagic on the basis of well-established CT and MRI criteria. In all these cases, hemorrhage was most obvious on GE scans as focal areas of marked hypointensity, including 10 cases in which SE MRI and CT demonstrated subtle, equivocal, or no evidence of hemorrhage. In addition, GE MRI allowed for the exclusion of hemorrhage in five cases with equivocal findings of hemorrhage on long TR SE MRI. GE MRI is a valuable adjunct to SE sequences for the detection or exclusion of hemorrhage in acute infarcts with equivocal or mild intensity changes on SE MRI. With further improvements in fast-scanning techniques and a greater understanding of the pathophysiology and clinical implications of hemorrhagic infarction, it may be possible to replace CT with a combination of SE and GE MRI as the primary imaging modality in the evaluation of acute infarction.