Background. Until recently, automated analysis of the cellular composition of bone marrow has not been possible, owing to problems in erythroblast identification, contamination by fat particles, and the heterogeneity of cell types. The XE-2100 automated blood cell counter can effectively count erythroblasts and granulocytes using laser light flow cytometry and DNA fluorochromes. In this study we have evaluated the performance of the XE-2100 for analyzing bone marrow cells. Material and Methods: The XE-2100 hematology analyzer performs white cell differential counts by measuring radiofrequency, direct current resistance, forward and side scatter of laser light, and fluorescence. Erythroblasts are identified by means of their specific fluorescent labelling. We have assessed i) the imprecision of the XE-2100 nucleated red blood cell (NRBC) counting using linear regression and the National Committee for Clinical Laboratory Standards (NCCLS) document H20A method of differences between duplicates, and ii) the comparability of NRBC count with microscopy counts using linear regression and analysis of differences. Results: Reproducibility of the XE-2100 measurements was excellent with coefficients of variation below 15%. We compared bone marrow granulocyte and erythroblast proportions by the XE-2100 with microscopy counts using linear regression and analysis of differences on 207 samples: Correlation between XE-2100 and the microscope was excellent for granulocyte percentage but slightly lower for erythroblasts. There was a tendency for the XE-2100 to slightly underestimate erythroblast percentage in most samples with a mean difference (XE-2100 microscope) of -13.4%.