Bone marrow biopsies (244) performed with a Jamshidi needle were evaluated in 53 children with leukemia or aplastic anemia. Adequate specimens were obtained in 85%. Results of cellularity estimated by biopsy were compared to the cellularity of the aspirate versus volumetric determination of the myeloid-erythroid layer (buffy coat). A wide discrepancy was noted between marrow cellularity confirmed by biopsy versus the aspirate or buffy coat. The greatest variance was seen in the hypercellular or normocellular marrows, as estimated by biopsy, in which 39% were misinterpreted as moderately or severely hypocellular by aspirate. Volumetric measurement of buffy coat was least acceptable for estimating cellularity. Thus the biopsy has proved to be an important and reliable indicator of bone marrow cellularity.