Myocardial blood flow was analyzed by radioisotope-labeled microspheres and ultrafast computed tomography (CT) in 16 closed-chest, anesthetized dogs. The first set of 10 dogs had CT and microsphere measurements before and after chromonar-induced increases in myocardial blood flow. A second set of six dogs had flows measured at control and during temporary reductions in regional flow produced by balloon cuff occlusion of the left anterior descending coronary artery. All dogs had four-slice, 20-instance CT scans after injection of a medium bolus (0.35 ml/kg) of contrast medium into a femoral vein simultaneous with injection of microspheres into the left atrium. CT myocardial flow was calculated as the change in myocardial CT numbers divided by the area from a blood pool time-density curve. A wide range of myocardial blood flows was produced as determined by microspheres (0 to 6.7 ml/min/g). Global flow of the first set of dogs was shown to have excellent correlation (r = .95, n = 17) for a limited range (.4 less than X less than 1.4 ml/min/g) of flows. Regional flows of these measurements demonstrated less correlation (r = .63, n = 110) but extended the range of flow to 1.7 ml/min/g. At higher flows (greater than 2.5 ml/min/g) the correlation for global and regional flows was not significantly different than zero. Regional ischemic flow correlation extended the linear range of flow to 0 ml/min/g (r = .62, n = 17). These results show that CT can measure myocardial blood flow over a limited but clinically relevant range of flows defined as slightly above normal to ischemic. These results indicate that another preparation of CT flow measurement must be sought for quantification of myocardial perfusion values significantly above normal.