Rabbit plasma albumin was labelled with I131, injected intravenously, and measurements were made of the radioactivity in plasma, urine, and feces over many days. In some experiments plasma radioactivity was fractionated into I131-albumin activity and that of labelled breakdown products. Curves of these radioactivities were compared with corresponding curves predicted by four mathematical models. Each model included a vascular and extravascular albumin compartment in transfer equilibrium, a radioactive breakdown products compartment, and an excretion compartment; but model A supposed I131-albumin catabolism to occur within the vascular system, model B within the extravascular compartment, model C within both, and model D within a separate compartment receiving albumin for catabolism from the plasma. The experimental data were reasonably well predicted by models A and C. However, model D, though data were insufficient for its complete validation, gave the best predictions and agrees with present knowledge of albumin catabolism. Various methods for calculating the rate of albumin breakdown are discussed. When calculations are based solely on the plasma radioactivity data, identical rates are predicted by models A, C, and D. When, as a valuable independent method, catabolism is calculated from plasma and excreted radioactivities, an error (ordinarily small) is incurred unless account is taken of the rate of passage of I131-albumin to the breakdown sites, and of the rate of excretion of the radioactive breakdown products.