A novel 31P NMR method is described that is capable of determining rapid changes in the intracellular levels of various phosphorus-containing compounds in an isolated, perfused working rat heart. This technique involves the gating of 31P NMR measurements to a heart that is alternately perfused with a modified Krebs-Henseleit medium containing 10 mM pyruvate and equilibrated with either 95% O2/5% CO2 or 95% N2/5% CO2. The experimental design allows up to three NMR measurements to be made during a single O2/N2 perfusion cycle. When these measurements are repeated at different intervals during the cycle, rapid changes in metabolite levels can be determined. Preliminary studies have shown that hearts remain hemodynamically stable to the aerobic/anoxic perfusion cycle as judged by heart rate, peak systolic pressure, aortic output, and coronary flow for at least 80 min in the magnet when subjected to cycle times of 4.5-s O2 and 1.5-s N2 perfusions. NMR measurements made under these conditions showed that a transition from full aerobic perfusion to this cycle revealed a new steady state, with an increased inorganic phosphate level from 6% total observable phosphorus to 10% and a possibly significant decreased measurement of creatine phosphate level (from 35 to 31%). Comparison of individual NMR measurements made during this perfusion cycle shows apparent rapid cyclical variations in intracellular pH and the levels of Pi, ATP, and NAD(H). These changes, expressed as variations above and below mean values measured during the cycle, showed that (a) intracellular pH, as measured by the chemical shift of Pi, reversibly decreases by more than 0.1 pH unit within 0.5-1 s following maximal anoxic perfusion and (b) coincident with a decrease in intracellular pH, Pi levels increased by a maximum of 30-40% whereas ATP levels decreased by a maximum of 15-20%. The amount of total observable phosphorous detected during the cycle is essentially constant. Unexpectedly, creatine phosphate levels are most stable, indicating that their levels are being maintained at the expense of ATP. Also unexpected is the finding that NAD(H) levels varied from maximal to undetectable levels during the perfusion cycle. The current method of aerobic/anoxic perfusion is capable of resolving metabolic events much faster than previous NMR methods and yielding information that is unobtainable by any other technique.