Heparin of five commercially available brands was used to study the disappearance of heparin anticoagulant activity in normal humans. The drug was administered intravenously by bolus injection and by continuous infusion. Heparin anticoagulant activity was determined by two assays: a diluted activated partial thromboplastin time (APTT) and an assay based on inactivation of bovine factor Xa, using a clotting system. After a bolus injection, the data fitted neither single exponential nor zero-order clearance. In semilogarithmic plots, heparin anticoagulant activity disappeared according to a slightly convex curve almost always preceded by a rapid initial loss of heparin anticoagulant activity. This disappearance profile was observed with all heparin regardless of the brand or assay system. Heparin anticoagulant activity estimated by the APTT disappeared faster than heparin anticoagulant activity estimated by the anti-Xa activity in the first phase. As expected, higher anticoagulant levels with the anti-Xa assay than with the APTT were also found on continuous infusion in normals as well as in patients treated for deep vein thrombosis or pulmonary embolism. The experimental data suggested a model based on the combination of a saturable and a linear clearance mechanism. These experimental data provide reliable guidelines for adjustment of the dose of heparin in single patients.