According to the recommendations of the Dosimetry Task Group of the American Association of Physicists in Medicine, blood-derived estimates of the red marrow (RM) dose from radiolabeled monoclonal antibodies (MAbs) are valid only if the RM is devoid of any specific uptake. There is, therefore, a clear need for an alternative method for estimating the RM dose in patients receiving MAbs that target normal or abnormal (malignant) bone marrow elements. Radiolabeled LL2, an anti-B-cell murine MAb, targets normal B cells and malignant lymphoma cells in the RM. This may result in an increased radiation dose to the RM through neighboring targeted activity. We investigated whether imaging-based estimates of the RM dose, particularly using sacral scintigraphy, correlate with myelotoxicity in non-Hodgkin's lymphoma patients who received 131I-LL2. The sacrum-based RM dose (RMs) was estimated from sacral activity by assuming that 9.9% of the total adult RM is contained in the sacrum. The sacrum was not used if there was focally increased or decreased sacral uptake. Myelotoxicity was assessed based on Radiation Therapy Oncology Group criteria. Twelve of 21 non-Hodgkin's lymphoma patients treated had adequate imaging, dosimetry, and follow-up to evaluate myelotoxicity. Eight of these patients had diffusely increased RM uptake on their MAb scans. The average estimated RMs in the eight patients was 168 +/- 62 cGy (mean +/- SD) with only 50 mCi 131I-LL2. Six of these patients (75%) developed grade 3 or 4 myelotoxicity. In contrast, the average RMs in the four patients who did not have any enhanced uptake on their scans was 71 +/- 30 cGy (P < 0.02). None of these patients developed grade 3 or 4 toxicity. These results suggest that image-based estimates of the RM dose may be predictive of myelotoxicity and should be used in patients with diffuse RM uptake on their scans.