The status of ATP as a possible coronary vasodilator remains poorly understood. The onset of hypoxia induced a rapid and transient increase of the ATP concentration in the coronary effluent of the isolated perfused rat heart from 0.8 +/- 0.2 nM to the average peak value of 1.3 +/- 0.2 nM (P less than 0.01) at 2 +/- 0.5 min; at the same time the coronary flow increased 2-fold so that the rate of ATP release increased from 10.2 +/- 2.9 to 21.4 +/- 4.2 pmol/g/min (P less than 0.005). Hypoxia also produced a peak rate release of adenosine of 93 +/- 5 nM/g/min occurring only after the peak increase of coronary flow and also after the peak release of ATP; at peak coronary flow, however, the adenosine concentration was sufficient for vasodilation (0.31 +/- 0.19 microM). Peak release of ATP and of adenosine preceded that of lactate dehydrogenase. 10(-6) M adrenaline induced a rapid increase of coronary flow and release of ATP, the concentration of which rose from 0.9 +/- 0.3 nM to an average peak of 1.7 +/- 0.2 nM (P less than 0.01) at 2 +/- 0.3 min. The rate of increase of ATP in the coronary effluent paralleled the rate of early rise of coronary flow, yet adenosine had also risen to vasodilatory values (0.28 +/- 0.5 microM). The absolute changes in the measured concentrations of ATP in the coronary effluent were more variable and 1000 X less in concentration than those of adenosine. Hence coronary dilation could be explained by adenosine without involving ATP, although an additional vasodilatory role for ATP could not be excluded, especially in the early phases of vasodilation. In one condition, hypoxic K-arrested hearts, the increase in coronary flow could not be linked to release of either adenosine or ATP. The changes in concentrations of potential vasodilators measured in the coronary effluent do not necessarily reflect changes in the interstitial fluid.