The univariate analysis of prognostic factors showed the value of clinical variables (age, symptoms, lymph node enlargement, splenomegaly, hepatomegaly, clinical stage), haematological variables (haemoglobin, platelet count, leucocyte count, lymphocyte count, bone marrow involvement and biopsy pattern, cleaved lymphocytes, prolymphocytes, LDT) and biochemical variables (BUN, creatinine, calcium, phosphorus, uric acid and albumin). The multivariate analysis chose the combination of Rai's stage, age, cervical lymph node involvement, phosphorus and BUN. Two predictive models capable of separating appropriately low, intermediate and high risk groups were developed and validated. The results were compared with others in the literature. The interest of predictive models arising from multivariate analysis is stressed.