In post-myocardial infarction patients, three-dimensional structure of the infarct as well as infarct size are likely to be important factors affecting mortality, cardiac function, and arrhythmias. Current morphological methods for determining three-dimensional infarct structure in autopsied hearts are inexact and time consuming. The cardiac magnetic resonance imaging techniques used in living patients have shown potential in determining infarct size and structure but have limited resolution for morphometric postmortem studies. The recent development of magnetic resonance microscopy raises the possibility that three-dimensional infarct structure can be quantified at microscopic levels in autopsied hearts. The purpose of this study was to determine the ability of magnetic resonance imaging at different spatial resolutions to differentiate infarcted from noninfarcted myocardium. Magnetic resonance imaging was performed at 2.0 T on cross sections taken from 10 autopsied hearts containing old myocardial infarcts. T1 was derived from six images with repetition times (TRs) for each image ranging from 100 to 3200 milliseconds. T2 was derived from multi-echo images with echo times (TEs) ranging from 10 to 60 milliseconds. Resolution was approximately 400 x 400 microns in 2-mm-thick slices. Sites of infarcted and noninfarcted tissue were identified from histological sections taken from each slice, and the T1 and T2 values of these sites were obtained. Microscopic images were acquired with voxels of 100 x 100 x 625 microns, representing tissue volumes more than 1000-fold smaller than conventional clinical images. In all cases, T1 of infarcted tissue (459 +/- 266 milliseconds, mean +/- SD) was greater than that of noninfarcted tissue (272 +/- 163 milliseconds). Also, in all cases, T2 of infarcted tissue (49 +/- 14 milliseconds) was greater than that of noninfarcted tissue (35 +/- 8 milliseconds). T1 and T2 values for infarcted tissue are significantly different from those of noninfarcted tissue (P < .001). Based on these findings, it should be possible to develop techniques to perform three-dimensional imaging and quantitation of infarcts with a resolution of 400 microns or less. When volumetric three-dimensional imaging was performed using a T1-weighted sequence, the resulting 256(3) arrays supported isotropic resolution at 400 microns (voxel volume, 0.064 mm3). Subsequent volume rendering using a compositing algorithm clearly shows the infarcted areas in three dimensions. The techniques demonstrate the potential for quantitative three-dimensional cardiac morphometry using magnetic resonance imaging.