The thrombogenicity of prothrombin complex concentrates (PCCs) has been known as a risk factor since their first clinical use about 30 years ago. The development of in vivo models to define the thrombogenic components in PCCs was instrumental in providing a logical basis for selecting in vitro assays to screen for the distribution of such components during the manufacture of PCCs, and to minimize their appearance in the final product. Even so, these thrombogenic components are not completely removed, as shown in our canine nonstasis model of thrombogenicity: PCCs were still found to elicit a thrombogenic response, shown by increased fibrinopeptide A, fibrin(ogen) degradation products, activated partial thromboplastin time, and decreased fibrinogen and platelet counts when clinically relevant doses were used. The new generation of high-purity factor IX (HP-FIX) concentrates differs from PCCs because these products contain only negligible amounts of clotting factors other than factor IX, lower amounts of activated clotting factors, and, in products we have assayed, no coagulant-active phospholipids. When we infused a number of HP-FIX products in the canine nonstasis model, no thrombogenic response was observed at doses considerably greater than PCC doses that did elicit a response. Likewise, HP-FIX products were much less thrombogenic than PCCs when tested in small-animal stasis and nonstasis thrombogenicity models. Small-animal models are also useful for evaluating the role of factor IXa as a potential thrombogenic contaminant of concentrates and ensuring minimal amounts in the final product. The limitations associated with extrapolating in vivo model data will be shown to be minimal if ongoing clinical studies continue to demonstrate the low thrombogenic potential of HP-FIX concentrates in humans.