We have compared apparent myocardial blood flow (MBFapparent) indicated by 99mTc-teboroxime, 96Tc-sestamibi and 201TI to true MBF indicated by radiolabeled microscopheres using a technique for the co-injection of four radionuclides in the same animal. Studies were performed using rats in a single-pass model to obtain global MBF and using dogs in a multiple-pass model to determine regional MBF. To provide a wide range of MBF, adenosine was administered intravenously and the left anterior descending coronary artery was then ligated in the dogs, or hypercapnia was induced by decreasing respiratory frequency in the rats. The microsphere formula for determining MBF was applied to all agents. When MBFapparent was plotted as a function of true MBF, the ability of each agent to measure changes in true MBF was demonstrated by the proximity of the plotted function to the line of identity. For both the single and multiple-pass studies, statistical analysis of the nonlinear relationship between MBFapparent and true MBF showed that 201TI and 99mTc-teboroxime approximate true MBF better than 96Tc-sestamibi (p less than 0.001) under the conditions used in the present studies. In the single-pass studies, 99mTc-teboroxime approximated true MBF better than 201Tl (p less than 0.05), but in the multiple-pass experiments, 201Tl approximated true MBF better than 99mTc-teboroxime in only one dog (p less than 0.01) with no difference in the other two. Determination of the permeability-surface area product, PS, for each agent shows that the higher fidelity to true MBF obtained with 201Tl and 99mTc-teboroxime is related to substantially greater PS values for these agents relative to 96Tc-sestamibi.