Radioactive microspheres accurately measure total coronary blood flow (+/- 10%) if the heart contains at least 400 spheres. Precision of the microsphere method depends upon the number of spheres in the region of interest and the reference sample, not on the radioactivity. Reliable flow can be calculated from reference samples collected from any vessel (carotid or femoral artery or aorta) in the beating, working heart and during cardiopulmonary bypass if the sample contains a minimum of 400 spheres. Spheres of 9 mu are distributed in the heart like diffusible indicators, while spheres greater than 9 mu preferentially stream to the endocardium, causing subendocardial flow to be overestimated. This overestimate is lessened with 15 mu spheres and can be readily taken into account. The microsphere method provides information about flow (ml/100 gm tissue/min) and allows comparisons with other regions (endocardial/epicardial flow ratios) but does not provide information about the adequacy of regional flow or its distribution; this must be assessed by biochemical, histochemical, and functional analysis.