One-hundred and ninety-four adult patients with histologically proven stage III and IV non-Hodgkin's lymphomas, treated with CVP (cyclophosphamide, vincristine and prednisone) alternated with ABP (adriamycin, bleomycin and prednisone), were analyzed to test the validity of the clinicoprognostic correlation offered by the working formulation in comparison with the Rappaport and Kiel classifications. Actuarial overall survival at 5 yr showed a significant difference among the three prognostic subgroups of the working formulation (low grade, 53.3%; intermediate grade, 47.5%; high grade, 27.7%). Overall survival of favorable subgroups of the Rappaport and Kiel classifications was superior to that of unfavorable prognostic groups. The percentage of systemic symptoms and bulky disease increased in patients with low-grade compared to those with intermediate-grade and intermediate-to-high-grade malignancy. The achievement of complete remission was not related to any of the prognostic groups of the Working Formulation, and no difference could be detected within the various prognostic groups of the Rappaport and Kiel classifications. Within the diffuse histiocytic lymphomas of the Rappaport classification, two groups with a different prognostic outcome were evidenced by the working formulation (G, with an overall survival of 50%, and H, with an overall survival of 26.7% at 5 yr) and by the Kiel classification. The possibility of reporting results in the three different groups of the working formulation instead of two can be considered a step forward. Within the diffuse histiocytic histology, the working formulation allows separation, as does the Kiel classification, into two main different prognostic subgroups.