The argyrophilic nucleolar organizer regions (AgNORs) were analysed in bone marrow biopsies from 80 patients with multiple myeloma (MM) at presentation. The mean AgNOR number per MM cell (AgNOR counts) and their distribution within the nucleus (configuration) were assessed. AgNOR counts were significantly associated with several recognized prognostic factors: Durie and Salmon clinical staging system (p = 0.02), percentage of plasma cells (PCs) in aspirates (p = 0.01) and in bone marrow biopsies (p = 0.0000), pattern of bone marrow involvement (p = 0.0003), calcaemia (p = 0.0005) and creatininaemia (p = 0.0003). AgNOR counts were also associated with the degree of PC differentiation (p = 0.0000). A single central cluster of 2-3 large-sized AgNORs (configuration A) was evident in most G1 MM; one cluster of 4-5 medium-sized dots or two clusters of 2-4 dots (configuration B) were seen in most G2 MM; many small-sized, scattered dots were present in G3 MM (configuration C). AgNOR counts and configuration were related to the prognosis: in the univariate analysis, the 5 year survival rate was 7% for cases with > 4.5 AgNORs/cell and 46% for cases with < or = 4.5 AgNORs/cell (p = 0.01), 53% for configuration A, 12% for configuration B and 0% for configuration C (p = 0.0000). AgNOR counts (p = 0.02) and configuration (p = 0.000) were independent prognostic variables in the multivariate analysis. The AgNOR counts were significantly higher in "fulminant myeloma" than in less aggressive cases (p = 0.002). The plasma cell labelling index (LI%), evaluated in 44 MM patients, showed significant correlation with prognosis: the 5 year survival rate was 51% for LI% < or = 1 and 17% for LI% > 1 (p = 0.02). More than 70% of patients with low LI% had low AgNOR counts and more than 70% of patients with high LI% had high AgNOR counts (p = 0.007). AgNOR counts and configuration reflect the myeloma cell mass, the degree of differentiation and the kinetics of the myeloma cells. They offer an exact evaluation of the tumour characteristics and can be useful additional parameters for MM prognosis.