Assessment of bone marrow involvement by malignant plasma cells is an important element in the diagnosis and follow-up of patients with multiple myeloma and other plasma cell dyscrasias. Microscope-based differential counts of bone marrow aspirates are used as the primary method to evaluate bone marrow plasma cell percentages. However, multiple myeloma is often a focal process, a fact that impacts the accuracy and reliability of the results of bone marrow plasma cell percentages obtained by differential counts of bone marrow aspirate smears. Moreover, the interobserver and intraobserver reproducibility of counting bone marrow plasma cells microscopically has not been adequately tested. CD138 allows excellent assessment of plasma cell numbers and distribution in bone marrow biopsies. We compared estimates of plasma cell percentages in bone marrow aspirates and in hematoxylin-eosin- and CD138-stained bone marrow biopsy sections (CD138 sections) in 79 bone marrows from patients with multiple myeloma. There was a notable discrepancy in bone marrow plasma cell percentages using the different methods of observation. In particular, there was a relatively poor concordance of plasma cell percentage estimation between aspirate smears and CD138 sections. Estimates of plasma cell percentage using CD138 sections demonstrated the highest interobserver concordance. This observation was supported by computer-assisted image analysis. In addition, CD138 expression highlighted patterns of plasma cell infiltration indicative of neoplasia even in the absence of plasmacytosis. We conclude that examination of CD138 sections should be considered for routine use in the estimation of plasma cell load in the bone marrow.