Little is known of the functional capacity of coronary collaterals in humans without occlusive coronary artery disease. Therefore the authors measured peripheral coronary pressure (PCP) and retrograde flow (RF) from coronary arteries at aortic valve replacement in 7 patients without occlusive coronary artery disease. Using a T connection interposed in left (LCA) and right (RCA) coronary perfusion lines, data were obtained during brief proximal occlusion of each line. PCP was expressed as a fraction of perfusion pressure (PP), and collateral resistance (CR) was calculated as PP/RF. Median values were as follows: for LCA, RF (ml/min) 2.4, CR (mm Hg/ml/min) 48, PCP (mm Hg) 15, PCP/PP 0.19; and for RCA, RF 1.7, CR 64, PCP 16, PCP/PP 0.24. In contrast, previous studies of well collateralized distal segments of diseased coronary arteries revealed mean RF 15.7, CR 5.1 and PCP/PP 0.50. Under the conditions of study, all vessels interconnecting the nondiseased RCA and LCA delivered flow and pressure less readily than collaterals to a single distal segment of a diseased coronary artery. Thus, collaterals in patients without diseased coronary arteries have an extremely limited capacity to transmit either flow or pressure. The absolute values of RF were small relative to the muscle mass perfused by each coronary artery, suggesting that perfusion of only one coronary artery in man during operation may not provide substantial perfusion for large portions of myocardium. Comparison of performance of collaterals supplying atherosclerotic and nonatherosclerotic coronary arteries indicates that proximal occlusion may be an important factor stimulating enhancement of collateral function. Moreover, the nitroglycerin induced improvement in collateral function seen in patients with chronic occlusive coronary disease was not demonstrable in patients without coronary occlusion. Thus, coronary collaterals may acquire nitroglycerin responsiveness as a result of changes induced by chronic coronary occlusion.