Although bone marrow biopsy pattern (BMBP) has long been suggested to be an independent prognostic factor in chronic lymphocytic leukemia (CLL), conflicting reports continue to appear in the literature. To investigate this issue we retrospectively reviewed 70 CLL patients who had undergone bone marrow biopsy at the time of diagnosis in a multivariate Cox regression analysis together with other prognostic factors. There were 51 (72.8%) males and 19 (27.2%) females with a median age of 60 years (range, 38-77). The median follow-up time was 24 months (range, 1-76), and median survival was 44 months. Thirtyfour patients (48.6%) had diffuse and 36 patients (51.4%) had nondiffuse BMBP (14 nodular, 11 interstitial, and 11 mixed). The median survival for diffuse and nondiffuse BMBP groups were 17 and 53 months, respectively (P = 0.05). Sixteen patients (22.9%) had stage A, 28 (40.0%) stage B, and 26 (37.1%) stage C disease according to the Binet system, and four patients (5.7%) had low-risk, 39 (55.7%) intermediate-risk, and 27 (38.6%) high-risk disease according to the modified Rai staging system. The difference between the median survivals of patients in different stages was statistically significant (P < 0.0001). The BMBP and staging systems that are thought to be significant predictors of prognosis were used to build a multivariate Cox proportional hazard model. BMBP was not found to add additional information to the prognostic value of the staging systems. Our results underline two points: first, the significance of BMBP must be investigated in multivariate analysis including the stage, and second, BMBP is not a dynamic prognostic parameter, it is an index of tumor burden and does not add any prognostic information beyond that provided by clinical stage. (C) 1999 Wiley-Liss, Inc.