We have investigated the role of MgATP in the reaction catalyzed by nitrogenase from Azotobacter vinelandii. There is a rapid burst of ATP hydrolysis in the pre-steady-state reaction that occurs on the same time scale as the electron transfer from dinitrogenase reductase to dinitrogenase. This burst corresponds to two ATP's hydrolyzed per electron transferred between the two proteins. Two MgATP molecules are bound to dinitrogenase reductase with dissociation constants of 430 microM and 220 microM. Investigation of the effect of MgATP concentration on the pre-steady-state kinetics of electron transfer from dinitrogenase reductase to dinitrogenase showed that there are two MgATP's required for this reaction, and the Km values are 220 microM and 970 microM. These values are similar to the dissociation constants for MgATP from dinitrogenase reductase and indicate that electron transfer between the two proteins is substantially slower than the binding and dissociation of MgATP from dinitrogenase reductase. The Km values for MgATP in steady-state H2 evolution were 390 microM and 30 microM. The decrease in the value of the second Km indicates that a slow, irreversible step occurs after the electron transfer from dinitrogenase reductase to dinitrogenase. It is possible to predict quantitatively the steady-state kinetics from the pre-steady-state kinetics, and this shows that the MgATP dependence of electron transfer is sufficient to account for effects of MgATP concentration on the steady-state H2 evolution catalyzed by nitrogenase. The hydrolysis of two ATP molecules when an electron is transferred between the two proteins of the nitrogenase system is sufficient to account for all of the ATP hydrolysis occurring in the steady-state reaction. The simplified scheme proposed to account for the MgATP dependency of the nitrogenase reaction indicates that the only role of MgATP is in support of the electron transfer from dinitrogenase reductase to dinitrogenase.