The feasibility of using magnetic resonance (MR) imaging to estimate myocardial infarct size was explored in an in vitro model using only the inherent differences in contrast between infarcted and noninfarcted myocardium. Eight dogs underwent coronary occlusion; their hearts were removed 6 hours later. Estimates of T2 for normal and infarcted myocardium were derived from MR images. Infarct size was quantified anatomically using triphenyltetrazolium-chloride (TTC) staining and compared with MR estimates. The T2 values derived from the images clearly discriminated between infarcted (126 +/- 22 msec) and normal myocardium (88 +/- 10 msec, P less than .05), providing images with good contrast between normal and infarcted myocardium. Comparable differences in T2 values were also noted from spectrometric determinations. Estimates of infarct size by MR imaging compared well with TTC estimates (r = 0.98) over a wide range of infarct sizes from 3% to 29% of the left ventricular mass. These results suggest the potential for in vivo quantification of infarct size based on the inherent contrast difference between infarcted and normal myocardium.