The activation and inactivation of human factor VIII by thrombin have been investigated by the use of thrombin inhibitors. The addition of inhibitors to nonactivated factor VIII blocks activation by thrombin. In contrast, their addition to factor VIII activated with thrombin does not block inactivation, but causes an initial period of decay that is more rapid than in the absence of inhibitor. This effect was seen only with protease inhibitors that inhibit thrombin. After the initial decay, low levels of factor VIII coagulant activity persist in the presence of inhibitors, but an assay specific for activated factor VIII showed this to be largely a result of the persistence of nonactivated factor VIII. Only in the case of reversible inhibition is activated factor VIII present in this plateau phase. Possible mechanisms that would account for these observations were studied by iterative computer simulation of model reactions. Two classes were considered: (formula: see text). The experimental results are inconsistent with the first mechanism, which predicts that thrombin indicators should stabilize activated factor VIII (VIIIt). Alternative mechanisms were studied where activation is thrombin-dependent, but inactivation is a first-order reaction (mechanism 2). This family of mechanisms includes those where VIIIt is an VIII. thrombin complex. Simulation of the addition of thrombin inhibitors to such model systems shows the initial rapid decay of activity characteristic of the experimental observations and predicts qualitatively the different effects of reversible and irreversible inhibitors that are observed in the plateau phase. These results argue strongly against a two-cleavage model for the activation and inactivation of factor VII by thrombin and support a one-cleavage model in which inactivation is due to first-order decay. In addition, they provide a plausible mechanistic explanation for the fact that serine protease inhibitors appear to inhibit thrombin-activated factor VIII.