Individual patient response to radioimmunotherapy is influenced by each patient's tumor burden, antibody clearance kinetics and the antibody-antigen interaction. In hematologic malignancies, wherein antibody access to tumor-cell associated antigen is rapid, mathematical modeling may provide a quantitative basis for assessing the impact of patient variability on a particular therapeutic protocol. Compartmental modeling analysis of antibody pharmacokinetics from a Phase I trial of 131I-labeled monoclonal antibody, M195 (anti-CD33), was used to estimate tumor burden in cases of acute myelogenous leukemia and the absorbed dose in liver, spleen and red marrow. The suitability of a nonlinear, two-compartment model for simulating M195 distribution in leukemia patients was evaluated by comparing model predictions with patient measurements. The results demonstrate that for directly accessible, hematologically distributed tumor cells, a two-compartment model fits observed patient biodistribution data and may provide information regarding both total tumor burden and tumor burden in the liver, spleen and red marrow. The model also provides biodistribution information for absorbed dose calculations to tissues that are not directly sampled. Such information is important in determining the optimum therapeutic dose of radiolabeled antibody for a given patient.