A method is presented for calculating the small-scale dosimetry of 211At in red bone marrow using chord-length distributions obtained from digitized histological images. This study used histological samples of bone marrow from beagle dogs to convey morphological information about cell conglomerations within bone marrow. Two 211At activity distributions were considered within the extracellular fluid and the surface of red bone marrow cells. Results confirmed the influence of cell conglomeration and activity distribution in determining the microdosimetry of red bone marrow. Average S* values of 1.6 x 10(-9) and 1.90 x 10(-9) Gy g Bq(-1) s(-1) were calculated for activity distributions located within the extracellular fluid and the surface of red bone marrow cells, respectively. The cumulated activity required to reduce survival probability to 0.37 also was calculated as a function of cell sensitivity for both activity distributions. The activity distribution on the cell surface resulted in a higher cell-killing efficiency, requiring a lower activity concentration of approximately 25% when compared with activity located in the extracellular fluid. Of relevance to potential clinical studies with 211At, the probability for zero hits for red bone marrow cells was > 10% for cumulated activities of less than 5.5 x 10(8) Bq s g(-1) in bone marrow.