Andersson and colleagues (1) have conducted an important study of patients exposed to Thorotrast in which they compute leukemia risk estimates for a-particle irradiation of bone marrow. Among 1,003 subjects injected with an average of 18.7 ml of Thorotrast (thorium dioxide colloid) and followed for an average of 20.4 years, 22 leukemias occurred. For acute myelogenous leukemia (n = 16) and myelodysplastic syndrome (n = 7) taken together, an excess of 173 cases per 104 persons per gray (173 per 104 Gy) was reported. The excess absolute risk would be 8.48 per 104 persons per year per gray (8.48 per 104 PY Gy). These risks are three times higher than previously estimated (2, 3). The authors conclude that the relative hazard for a-particle exposure to bone marrow compared to X-ray exposure is lower than 20, and they reinforced the concern raised by others that assigning an RBE of 20 to low-LET risk coefficients to predict leukemia risks from high-LET exposures may not be appropriate (3, 4). Recent data on the incidence of leukemia in the atomic bomb survivors have become available to allow more comparable age-specific comparisons with the Thorotrast series (5). For survivors over the age of 20 when exposed, the absolute excess risk of leukemia after 37 years of observation is 3.06 per 104 PY Gy, or about 113 per 104 Gy (D. Preston, personal communication). The data for the atomic bomb survivors, however, do not include myelodysplastic syndromes (MDS) or chronic lymphocytic leukemia (CLL). A comparable risk coefficient from the Thorotrast study would combine all non-CLL leukemias (n = 20) and exclude the MDS. Assuming that the reduced number of cases would result in a proportionate reduction in the reported risk coefficient for AML and MDS, an approximate value of 150 per 104 Gy can be estimated. This cumulative risk is only 1.33 times greater than the comparable A-bomb estimate, and certainly closer to 1 than to 20. Several factors, however, might affect the accuracy of the risk coefficient for Thorotrast patients, such as the presence of Thorotrastomas (which would result in less radioactivity reaching the vascular system), and the percentage of Thorotrast assumed to reach the bone marrow. Thorotrast granulomas (Thorotrastomas) are common lesions found in patients given Thorotrast. They occur at the injection site and ar caused by Thorotrast deposits that do not reach the vascular system. On. the order of 3-5% of injected Thorotrast can remain at the injection site, and in some instances the percentage can be much higher. If an adjustment for Thorotrastomas was not made in the Danish series, the actual dose to bone marrow was probably somewhat lower than reported and the risk coefficient would then be somewhat higher than computed. Bone marrow dose also depends on the amount of injected volume that is taken up by the macrophages in the bone marrow, as well as the amount of a-particle self-absorption within Thorotrast aggregates. Andersson et al. (1) used the determination of Kaul and Noffz (6) that,9% of the injected thorium would accumulate in the red bone marrow. They also reference a Japanese study that found the fraction of Thorotrast in the bone marrow to be as high as 21% (7) and a recent whole-body autopsy study that reported a 32% distribution of Thorotrast in bone (8). For the whole-body autopsy evaluation, the percentage uptake in the bone marrow, as opposed to total bone, was actually 23%, close to the Japanese value. Assuming 20% uptake rather than 9% might have resulted in a higher estimated dose to marrow, and an associated lowering of the risk coefficient.