Although marrow suppression is usually the dose-limiting toxicity in non-marrow ablative radionuclide therapy, calculated marrow dose has rarely been used for prescribing the radioactivity to be administered. This study assesses the correlation of myelotoxicity with mCi/m(2), patient-specific lean body dose, marrow dose from blood and body of reference man, or from blood and body using the patient-specific mass. Fourteen prostate cancer patients were treated with (131)I-CC49. Radioactivity in blood and body was determined and used to calculate their contributions to the marrow dose. Platelet nadir expressed as percentage (%) of the initial baseline was used as an indicator for myelotoxicity. Correlation between platelet nadir (%) and myelotoxicity predictors was evaluated. Platelet nadirs (%) varied substantially (5-33%) for a small range of injected radioactivity/m(2) (68-78 mCi/m(2), 2.5-2.9 GBq/m(2)). Patient-specific total body dose based on lean body mass exhibited a weak correlation (r = 0.48) with platelet nadir. Marrow dose from blood and body of reference man had a better correlation (r = 0.73). Patient-specific marrow dose from blood and body (or lean body) had a similar correlation (r = 0.74 or 0.73). Radioactivity in the remainder of the body contributed only 28% of the total dose, and thus changes to this dose component had small impact on total marrow dose. Marrow dose was a better predictor for myelotoxicity than mCi/m(2) or lean total body dose in this non-marrow targeting (131)I-antibody therapy with high blood contributions to total dose.