Reported values of the dissociation constant, K(d), of ATP with the E1 conformation of the Na(+),K(+)-ATPase fall in two distinct ranges depending on how it is measured. Equilibrium binding studies yield values of 0.1-0.6 microM, whereas presteady-state kinetic studies yield values of 3-14 microM. It is unacceptable that K(d) varies with the experimental method of its determination. Using simulations of the expected equilibrium behavior for different binding models based on thermodynamic data obtained from isothermal titration calorimetry we show that this apparent discrepancy can be explained in part by the presence in presteady-state kinetic studies of excess Mg(2+) ions, which compete with the enzyme for the available ATP. Another important contributing factor is an inaccurate assumption in the majority of presteady-state kinetic studies of a rapid relaxation of the ATP binding reaction on the time scale of the subsequent phosphorylation. However, these two factors alone are insufficient to explain the previously observed presteady-state kinetic behavior. In addition one must assume that there are two E1-ATP binding equilibria. Because crystal structures of P-type ATPases indicate only a single bound ATP per alpha-subunit, the only explanation consistent with both crystal structural and kinetic data is that the enzyme exists as an (alphabeta)(2) diprotomer, with protein-protein interactions between adjacent alpha-subunits producing two ATP affinities. We propose that in equilibrium measurements the measured K(d) is due to binding of ATP to one alpha-subunit, whereas in presteady-state kinetic studies, the measured apparent K(d) is due to the binding of ATP to both alpha-subunits within the diprotomer.