1. A working rat heart preparation was used to study the release of adenosine-5'-triphosphate (ATP) into the coronary sinus effluent in response to hypoxia. 2. The left ventricle was set to pump against an hydrostatic pressure of 65 cm water; the left atrial filling pressure was kept constant at 10 cm water. The power output of the heart at these pressures was estimated to be approximately one half of the maximum power development. 3. Samples for ATP assay were collected (a) 30 sec before onset of hypoxia, (b) 60-90 sec after onset of hypoxia, (c) 5 min after restoration of oxygenated buffer solution. Respective concentrations of ATP were (nM +/- S.E.) 0.63 (+/- 0.18), 4.70 (+/- 0.39) and 0.63 (+/- 0.06). The total amounts of ATP detected were (p-mole/min) 5.9 (+/- 0.9), 46.1 (+/- 6.0) and 5.5 (+/- 1.2) respectively. 4. Viability of the hearts was judged to be satisfactory on the following grounds. Alterations in left atrial filling pressure produced typical Frank-Starling responses of the left ventricle. Oxygen extraction from the perfusate increased in response to increased workload. Coronary blood flow increased immediately upon introduction of hypoxic conditions and mechanical recovery from hypoxia was always complete within 5 min of restoring oxygen. 5. In view of the marked extracellular ATPase activity it is concluded that significant vasodilatory concentrations of ATP are released into the myocardial extracellular space in response to hypoxia. A scheme is proposed describing the possible role of adenine nucleotides in the local control of myocardial blood flow.