In radiolabeled antibody therapy, imaging and biopsy-based methods are used to estimate marrow activity concentration when the administered antibody localizes to the marrow. Absorbed dose estimates obtained using such measurements may be subject to large variability due to the potential for regional differences in marrow activity concentration. This variability was examined in ten patients with leukemia after administration of 131I-labeled HuM195 antibody.Regions of interest were drawn around the head and neck of the humerus and femur (both sides) and around lumbar vertebra 3 (L3) and 4 (L4) on a series of planar images collected at multiple times postadministration of the antibody. A single exponential fit to each attenuation-corrected, time-activity curve was obtained to estimate clearance half-life and the back-extrapolated percent injected activity.The activity concentration in the femoral head and neck (mean and s.d. = 0.04 +/- 0.02 %ID/g) was not significantly different than that measured in L3 and L4 (0.06 +/- 0.02% ID/g) but was not significantly lower than the concentrations measured in the humeral head and neck regions (0.07 +/- 0.03 %ID/g, p < 0.05). Although half-life estimates differing by more than a factor of 2 were observed in half-life between regions overall. S-factors were used for individual marrow regions to determine the mean absorbed dose to marrow in the femoral and humeral heads and the lumbar vertebrae (L3 and L4) which were 0.66 +/- 0.3, 1.0 +/- 0.3 and 2.2 +/- 0.5 mGy/MBq (2.4, 3.8 and 8.3 rad/mCi), respectively.A single value is generally quoted for the absorbed dose delivered to the red marrow following marrow-localizing radiolabeled antibody administration. These results suggest that the regional marrow dose may differ significantly from the mean.