In the 24 hours after coronary occlusion, some defects on thallium-201 images decrease in size. This study examined the mechanism of the decrease in defect size in dogs with experimental coronary occlusion. The left anterior descending coronary artery was permanently occluded in chronically instrumented awake dogs. Myocardial blood flow was measured with radioactive microspheres immediately before occlusion, 30 minutes, and 24 hours after occlusion. Thallium-201 was injected and imaging was performed 2 to 6 days before occlusion, 30 minutes, and 24 hours after occlusion. Two dogs, in which less than 1% of the left ventricle was infarcted, had no defects on the 30-minute postocclusion images. In four dogs the thallium-201 images did not change appreciably over 24 hours (group 1), while in three dogs the image defect size decreased (group 2). In groups 1 and 2 respectively the percent increase in blood flow to the hypoperfused area was: subendocardium 9 ± 7% vs 31 ± 15%; subepicardium 26 ± 6% vs 47 ± 2% (p < 0.05). Therefore a decrease in the size of thallium-201 image defects occurred if a sufficient increase in collateral flow to the hypoperfused area took place in the first 24 hours after coronary occlusion. The defects appeared to decrease in size even in the absence of a detectable change in the lateral borders of the hypoperfused area.