Usually, the red marrow (RM) is the first dose-limiting organ in radioimmunotherapy. However, several studies have obtained only poor correlations between the marrow doses and the resulting toxicities. Furthermore, RM doses are mostly not determined directly but are derived from blood doses by assuming a ratio that is, over time for the respective conjugates, more or less constant between blood and marrow activities. The aim of this study was to determine, in a mouse model, this RM:blood activity ratio for various immunoconjugates, to investigate whether there may be differences between complete IgG and its fragments with various labels ((125/131)I versus (111)In, (88/90)Y, or 213Bi), and to analyze, in more detail, factors other than just total dose, such as dose rate or relative biological effectiveness factors, that may influence the resulting myelotoxicity. The maximum tolerated activities (MTAs) and doses (MTDs) of several murine, chimeric, and humanized immunoconjugates as complete IgG or fragments (F(ab)2 and Fab), labeled with beta(-)-emitters (such as 131I or 90Y), Auger electron-emitters (such as 125I or (111)In), or alpha-emitters (such as 213Bi) were determined in nude mice. Blood counts were monitored at weekly intervals; bone marrow transplantation was performed to support the assumption of the RM as dose-limiting. The radiation dosimetry was derived from biodistribution data of the various conjugates, accounting for cross-organ radiation; besides the major organs, the activities in the blood and bone marrow (and bone) were determined over time. Whereas no significant differences were found for the RM:blood ratios between various IgG subtypes, different radiolabels or various time points, differences were found between IgG and bi- or monovalent fragments: typically, the RM:blood ratios were approximately 0.4 for IgG, 0.8 for F(ab')2, and 1.0 for Fab'. Nevertheless, at the respective MTAs, the RM doses differed significantly between the three conjugates: e.g., with 131I-labeled conjugates, the maximum tolerated activities were 260 microCi for IgG, 1200 microCi for F(ab)2, and 3 mCi for Fab, corresponding to blood doses of 17, 9, and 4 Gy, respectively. However, initial dose rates were 10 times higher with Fab as compared to IgG, and still 3 times higher as compared to F(ab)2; interestingly, all three deliver approximately 4 Gy within the first 24 h. The MTDs of all three conjugates were increased by BMT by approximately 30%. Similar observations were made for 90Y-conjugates. Higher RM doses were tolerated with Auger-emitters than with conventional beta(-)-emitters, whereas the MTDs were similar between alpha- and beta(-)-emitters. In accordance to dose rates never exceeding those occurring at the single injection MTA, two subsequent injections of two doses of 80% of the single shot MTA of 131I- or 90Y-labeled Fab' and two doses of 100% of the single shot MTA of 213Bi-labeled Fab' were tolerated without increased lethality, if administered 24-48 h apart. In contrast, reinjection of bivalent conjugates was not possible within 6 weeks. These data suggest that the RM:blood activity ratios differ between IgG and fragments, although there is no anatomical or physiological explanation for this phenomenon at this point. In contrast to the current opinion, indication for a strong influence of the dose rate (or dose per unit time), not only total dose, on the resulting toxicity is provided, whereas the influence of high-linear energy transfer (alpha and Auger/conversion electrons) versus low-linear energy transfer (beta and gamma) type radiation seems to be much lower than expected from previous in vitro data. The lower myelotoxicity of Auger-emitters is probably due to the short path length of their low-energy electrons, which cannot reach the nuclear DNA if the antibody is not internalized into the stem cells of the RM.