We determined the effects of time, type of anesthesia, and myocardial infarction on loss of radioactive microspheres averaging 9 or 15 micrometers diameter from left ventricular myocardium. The principle used to compute losses was comparison of the number of microspheres injected directly into coronary arteries to the numbers remaining in myocardium, appearing within 2-4 min in the coronary sinus, or trapped in the lungs. Losses of 9-micrometers microspheres within 2 min of injection were significantly greater for halothane (mean 6.3%) than nitrous oxide anesthesia (mean 3.3%), and in the next 2 h increased to 11.7 and 7.9%, respectively. Over 5 wk in conscious dogs losses were as high as 40 and 11% for 9 and 15 micrometers microspheres, respectively. Losses were not greater for infarcted than normal muscle, and negligible radioactivity appeared in paracardiac lymph nodes. Microspheres leaving the heart were almost all below 10.3 micrometers diameter, so that microspheres with diameters 10-14 (mean 12) micrometers might be the best size to use for myocardial studies.