A histomorphometric (planimetric) study was performed on trephine biopsies of the bone marrow taken at presentation from 65 patients (31 males and 34 females, with a median age of 48 years) with chronic myeloid leukemia (CML). Specimens from 20 patients (9 males and 11 females, with a median age of 53 years) without any hematologic disorders served as controls. Of the various histologic variables tested, only the counts of neutrophilic granulocytes per 1 sq mm, the ratio of granulocytopoiesis to megakaryopoiesis and the density of reticulin (argyrophilic) fibers revealed a significant correlation with the prognosis. The CML patients were separated into two groups with different survival patterns. Group I (34 patients with a median survival of 24 months) mostly contained cases with the so-called "megakaryocytic subtype" of CML, which is accompanied by variable degrees of fibrosis; group II (31 patients with a median survival of 36 months) mainly contained cases with the "granulocytic subtype," which is not accompanied by myelofibrosis. Among the morphometric parameters, a positive correlation existed between the megakaryocyte count and the reticulin fiber density, which underlines the important role of that cell lineage in fibrillogenesis. There were multiple interrelationships between the histomorphometric variables and the laboratory data. Consequently, multivariate regression methods (using Cox's proportional hazards model) were applied to assess the relative predictive value of the patient characteristics for survival. The derived prognostic model divided the patients into two risk groups, with median survivals of 14 and 41 months, respectively. In order of their entry into the regression model, these variables were percentage of neutrophils in the differential blood count, amount of granulopoiesis, liver size, percentage of peripheral myeloblasts and density of reticulin fibers in the bone marrow. In comparing the two patient groups, based on bone marrow histomorphometric parameters, this model revealed that two of those factors (amount of granulopoiesis and density of reticulin fibers) had a significant correlation with the prognosis.