To assess the tumor targeting, safety, and efficacy of monoclonal antibody 131I-labeled CC49 in patients with androgen-independent prostate cancer, 16 patients received 75 mCi/m2 of the radiolabeled antibody after 7 days of IFN-gamma pretreatment. Sequential tumor biopsies in three patients showed a median 5-fold (range, 2-6-fold) increase in the proportion of cells staining positively for the TAG-72 antigen, whereas one showed a decrease in staining. Fourteen patients received 131I-labeled CC49, whereas 2 showed a disease-related decrease in performance status, precluding antibody treatment. The antibody localized to sites of metastatic androgen-independent prostate cancer in 86% (12 of 14; 95% confidence interval, 57-95%) of cases. Both osseous and extraosseous sites were visualized, and in six (42%) patients, more areas were visible when the radioimmunoconjugate was used than were apparent when conventional scanning techniques were used. The localization of the conjugate in the marrow cavity was usually a site not visualized by the radionuclide bone scan, in which the isotope localizes primarily to the tumor-bone interface. The dose-limiting toxicity was thrombocytopenia because five (36%) patients showed grade IV and seven (50%) showed grade III effects. In addition, six (42%) patients, four of whom were hospitalized, showed a flare in baseline pain, and four showed a decrease in pain. No patient showed a >50% decline in prostate-specific antigen, although radionuclide bone scans remained stable in four cases for a median of 4 months. The results are consistent with dosimetry estimates showing that the delivered dose to tumor was subtherapeutic and suggest that approaches that exclusively target the bone tumor interface or the marrow stroma may be unable to completely eradicate disease in the marrow cavity. For CC49, improving outcomes would require repetitive dosing, which was precluded by the rapid development of a human antimouse antibody response.