Thrombin plays a central role in venous and arterial thrombosis. We utilized two different rabbit models of in vivo thrombosis to investigate the effect of inhibitors of thrombin generation and thrombin activity. The agents tested were specific inhibitors of factor Xa (fXa) [N2-[(phenylmethyl)sulfonyl]-D-arginyl-N-[(1S)-4-[(aminoiminomethyl++ +)a mino]-1-(2-thiazolylcarbonyl)butyl]-glycinamide (C921-78)] and thrombin [D-phenylalanyl-N-[4-[(aminoiminomethyl)amino]-1-(chloroacetyl)but yl]-L-prolinamide (PPACK)], as well as drugs that affect both thrombin and fXa, unfractionated and low molecular weight (enoxaparin) heparin. The agents administered as constant intravenous infusion were evaluated for antithrombotic efficacy in anesthetized rabbits. All four agents were capable of dose dependent inhibition of thrombosis in venous and arteriovenous thrombosis models. However, due to the more aggressive nature of thrombotic stimulation in the arteriovenous shunt model, complete cessation of thrombus growth was not achieved for any of the agents at the doses tested. Comparison between the agents focused on the differences in extension of coagulation parameters (activated partial thromboplastin time, prothrombin time, thrombin clotting time), changes in hematological parameters, and extension of rabbit cuticle bleeding time at doses required to produce maximum inhibition in the thrombosis models. In the venous thrombosis model at the maximally effective dose, C921-78 had minimal extension of ex vivo clotting parameters, while enoxaparin and unfractionated heparin demonstrated a two to sevenfold increase in activated partial thromboplastin times, and PPACK had a threefold extension of thrombin clotting times. In addition, unlike the other three agents, which exhibited no significant changes in hematological parameters, PPACK demonstrated dose dependent thrombocytopenia. A standardized cuticle bleeding time was used as a measure of perturbation of hemostasis. The agents were evaluated for significant increases in bleeding time at doses up to eight times that needed to completely inhibit venous thrombus formation. Unfractionated heparin displayed a significant bleeding time effect at the dose required to inhibit venous thrombosis (100 u/kg+2 u/kg/min). Enoxaparin and PPACK caused significant bleeding time extensions at four times the fully efficacious venous dose (800 u/kg+8 u/kg/min and 30 microg/kg/min). By contrast, C921-78 did not significantly increase bleeding time even at eight times the maximally effective dose (240 microg/kg+7.2 microg/kg/min). Our results demonstrate that specific inhibition of fXa can be utilized to derive potent antithrombotic activity without disrupting extravascular hemostasis.