The authors previously developed a filter device containing immobilized protamine (termed "protamine filter") that could be used to remove heparin during extracorporeal perfusion. In vivo studies involving dogs showed that the protamine filter removed more than 50% of heparin from the animals' blood circuit in less than 20 min. In addition, the use of the protamine filter did not elicit statistically significant protamine induced hemodynamic and thrombocytopenic responses. Biocompatibility of the protamine filter was also evaluated, with the focus on its effect on the coagulation cascade, the complement system, and the blood antithrombin III levels. Results showed that heparin adsorbed to the protamine coated surface retained 20% of its original activated partial thromboplastin time activity, rendering the coated surface antithrombotic. Activation of the coagulation system by the protamine coated membrane and the untreated cellulose membrane, as measured by the elevation of prothrombin fragment F1 + 2 levels, was statistically identical. The CH50 hemolytic assay showed that the protamine coated membrane produced a reduction of 1.2 +/- 0.8% of the total complement levels, as compared to 9.4 +/- 1.6% by the untreated membrane. In addition, the change in C3a des Arginine levels after 30 min of circulation was 1.5 +/- 0.2 mg/ml by the protamine filter, as compared to 2.1 +/- 0.1 mg/ml by the untreated membrane. Unlike native heparin that would bind with antithrombin, heparin adsorbed on the protamine coated surface was devoid of such activity, and produced no depletion of circulating antithrombin. Because of the limited capacity of the protamine filter, the future system is envisioned to consist of two filters; while one filter is removing heparin the other will be regenerated. With a recently developed heparin sensor, it should be possible to design a sensor directed, biofeedback, two filter heparin removal system.