Rabbit antithrombin III (AT), purified by heparin-agarose, was labeled with iodine-131 by either the glucose oxidase-lactoperoxidase or iodine monochloride techniques. When intravenously injected, the disappearance of the 131I-AT from plasma was characterized by rapid initial decreases, and three-exponential equations were required for best fit of the plasma disappearance curves. This rapid 131I-AT removal was not caused by denaturation, as shown by comparison with results obtained when 131I-AT was biologically screened (injected into a first rabbit, and then transferred 16 h later in whole plasma to a second for kinetic evaluation) before injection. Thus, the same rapid initial loss of plasma 131I-AT was observed with screened preparations, and the plasma fractional catabolic rates of 0.716 +/- 0.048 and 0.673 +/- 0.051 day-1 for unscreened and screened 131I-AT were not significantly different. These results support the hypothesis that a vascular-endothelial AT compartment is present in rabbit. The fractions of the total-body AT in the plasma, the vascular-endothelial and the extravascular compartments were 0.337 +/- 0.031, 0.178 +/- 0.056, and 0.485 +/- 0.069, respectively. Two three-compartment kinetic models are discussed. The first pictures AT as distributing independently between plasma and two other compartments, and the second sees AT as first passing to the vascular-endothelial compartment, and then directly into the extravascular compartment. The plasma 131I-AT kinetic data was consistent with both models, but the sizes of the vascular-endothelial compartments were best predicted by the second. If AT catabolism was assigned to the plasma, both models generally underpredicted the whole-body radioactivities, while assignment of breakdown to the extravascular compartment generally resulted in overpredictions. This suggests that AT catabolism occurs from both plasma and extravascular compartments.