Forward [creatine phosphate (CP) → adenosine 5'-triphosphate (ATP)] and reverse (ATP → CP) fluxes of myocardial creatine kinase (CK) measured by using 31P nuclear magnetic resonance (NMR) and conventional saturation transfer (CST) methods are unequal; this is a paradoxical result because during steady state fluxes into and out of the CP pool must be the same. These measurements, however, treat the CK reaction as a two-site exchange problem and ignore the presence of the ATP ⇆ Pi exchange involving the ATPases. We have applied a method [Ugurbil, K. (1985) J. Magn. Reson. 64, 207] based on the saturation of multiple resonances, by which a single unidirectional rate constant can be measured unequivocally in the presence of multiple exchanges, to the measurement of CK fluxes in isovolumic rat hearts perfused under three different conditions; two of the three perfusion conditions showed a large discrepancy in the CK fluxes determined by CST, and one did not. In contrast, when the effect of the ATP ⇄ Pi exchange on the CK rate measurements was eliminated, multiple saturation transfer (MST) measurements on the same hearts yielded equal forward and reverse fluxes in all cases. The rate constant for the ATP → CP conversion measured by MST was larger than the value obtained by the conventional methodology whereas both methods gave the same rate constant in the CP → ATP direction. These results demonstrate that the cause of the paradoxical data obtained by CST measurements of CK kinetics is the ATP Pi exchange and that CK rates when determined rigorously are consistent with the CK reaction being in equilibrium. Comparison of the MST and CST data suggests that in the myocardium, a three-site CP ⇄ ATP ⇆Pi scheme adequately but not fully accounts for the phosphate exchange among these compounds. The magnitudes of the CK fluxes measured by MST and CST were ~ 2-fold larger in hearts perfused with glucose in the absence of insulin compared to hearts where the perfusate was supplemented with pyruvate or insulin. The higher CK rate is probably the reason why the discrepancy in CK fluxes as measured by CST is not prominent in glucose-perfused hearts; this follows from the fact that for a given rate of ATP ⇄ Pi interconversion, the confounding influence of this exchange on the CST measurements of CK should diminish as CK fluxes increase.