We investigated the role of myoglobin (Mb) in supplying O2 to mitochondria during transitions in cardiac workload. Isovolumic rabbit hearts (n = 7) were perfused retrogradely with hemoglobin-free Tyrode solution at 37 degrees C. Coronary venous O2 tension was measured polarographically, and tissue oxygenation was measured with two-wavelength near-infrared spectroscopy (NIRS), both at a time resolution of approximately 2 s. During transitions to anoxia, 68 +/- 2% (SE) of the NIRS signal was due to Mb and the rest to cytochrome oxidase. For heart rate steps from 120 to 190 or 220 beats/min, the NIRS signal decreased significantly by 6.9 +/- 1.3 or 11.1 +/- 2.1% of the full scale, respectively, with response times of 11.0 +/- 0.8 or 9.1 +/- 0.5 s, respectively. The response time of end-capillary O2 concentration ([O2]), estimated from the venous [O2], was 8.6 +/- 0.8 s for 190 beats/min (P < 0.05 vs. NIRS time) or 8.5 +/- 0.9 s for 220 beats/min (P > 0.05). The mean response times of mitochondrial O2 consumption (VO2) were 3.7 +/- 0.7 and 3.6 +/- 0.6 s, respectively. The deoxygenation of oxymyoglobin (MbO2) accounted for only 12-13% of the total decrease in tissue O2, with the rest being physically dissolved O2. During 11% reductions in perfusion flow at 220 beats/min, Mb was 1.5 +/- 0.4% deoxygenated (P < 0.05), despite the high venous PO2 of 377 +/- 17 mmHg, indicating metabolism-perfusion mismatch. We conclude that the contribution of MbO2 to the increase of VO2 during heart rate steps in saline-perfused hearts was small and slow compared with that of physically dissolved O2.