The kinetic mechanism of the NADP-dependent isocitrate dehydrogenase of Escherichia coli was investigated using initial steady-state kinetic analyses. Kinetic coefficients, obtained using natural and alternative substrates with the wild-type and two mutant enzymes (S113L and S113N), suggest that the forward reaction [the oxidative decarboxylation of (2R, 3S)-isocitrate by NADP] of the wild-type enzyme is a steady-state random mechanism, with catalysis more rapid than product release. The mechanism of the wild-type enzyme becomes rapid-equilibrium random when an alternative substrate [(2R)-malate or NAD] is used. The mutant enzymes always display rapid-equilibrium random kinetics, and for each enzyme the apparent dissociation constant of each substrate from the binary complex [Kia = E·A/(EA)] is similar to its apparent dissociation constant from the Michaelis complex [Ka = (EB)·A/(EAB)], which suggests that the binding of one substrate is independent of the binding of the second. When the wild-type enzyme catalyzes the forward reaction, the apparent dissociation constant, KiISo, is equal to its equilibrium dissociation constant, KdIso, determined from equilibrium binding studies. However, the apparent dissociation constant of the cofactor, KiNADP, is far smaller than its equilibrium dissociation constant, KdNADP. This is consistent with the proposed mechanism, because simulations show that when catalysis is steady-state and product release is rate-limiting, KiNADP and KNADP will be far smaller than KdNADP, while KiIso and KIso remain similar to KdIso. Product inhibition studies support the steady-state random mechanism of the wild-type enzyme. The rapid-equilibrium random mechanisms of the mutant enzymes provide evidence for the existence of E·Iso·NADPH and E·αKg·NADP abortive complexes and demonstrate that α-ketoglutarate and NADPH each bind to the free enzyme. Initial steady-state rate and product inhibition studies of the reverse reaction indicate a random addition of α-ketoglutarate and NADPH, with CO2 possibly binding last. Dead-end inhibition studies, using tricarballylate (propane-1,2,3-tricarboxylic acid) as an analog of isocitrate and adenosine 2′,5′-diphosphate as an analog of NADP, are incompatible with ordered mechanisms in either direction.