In 1978, Cancer and Leukemia Group B initiated a randomized study to determine the usefulness of the addition of bleomycin and/or high-dose methotrexate to standard therapy for the treatment of certain adult non-Hodgkin's lymphomas. Between 1978 and 1985, 177 patients with diffuse large cell lymphoma (DLCL) and 97 patients with other intermediate-grade non-Hodgkin's lymphoma were randomized to receive therapy with three courses of cyclophosphamide, adriamycin, vincristine, and prednisone (CHOP) every 3 weeks with or without low-dose bleomycin by continuous IV infusion. Responders after three courses were further randomized to 3 weeks of therapy with either high-dose methotrexate (3 gm/m2/week intravenously with leucovorin rescue) or standard-dose methotrexate (30 mg/m2/week orally without rescue). Therapy was concluded with three additional courses of CHOP. Neither the addition of low-dose infusion bleomycin nor the use of high-dose rather than low-dose methotrexate had significant effects on response for patients with DLCL; complete response rates for the four treatment programs ranged from 47% to 51%. Median failure-free survival (FFS) for the entire group of DLCL patients was 12 months; 5-year FFS was 27%. There was no significant effect on FFS from the addition of either low-dose bleomycin to CHOP (5-year FFS: CHOP, 28%; CHOP-B, 26%, P = 0.81), or from the use of different doses of methotrexate (5-year FFS: high-dose, 34%; standard-dose, 33%, P = 0.51). Patients with follicular large cell lymphoma, with or without diffuse areas, had a better FFS (5-year FFS, 47%) than patients with DLCL (5-year FFS, 27%), while the patients with the other histopathologic subtypes of diffuse lymphomas had the poorest FFS (5-year FFS, 16%).