The French-Italian Cooperative Study Group included patients with poor-prognosis AIDS-related non-Hodgkin's lymphoma (NHL), defined as those with performance status (PS) > or = 3 and/or opportunistic infections (OI), in a prospective study with a 50% reduced-dose combination chemotherapy regimen: CHVmP-Vincristine-bleo (cyclophosphamide 300 mg/m2 i.v. day 1, doxorubicin 25 mg/m2 i.v. day 1, teniposide 30 mg/m2 i.v. day 1, prednisone 20 mg/m2 per os days 1-5, vincristine 2 mg i.v. day 15, and bleomycin 10 mg i.v. day 15), given every 21 days for eight cycles, and concomitant zidovudine 500 mg per os per day. The aims of this combined treatment were to reduce bone marrow toxicity and infectious complications related to chemotherapy (with a low-dose chemotherapy regimen), and to control the HIV and related infectious complications (with zidovudine therapy). Thirty-seven patients entered this prospective study. At the time of the NHL diagnosis, 41% of the patients had asymptomatic HIV infection, 27% had ARC and 32% had already had CDC-defined diagnoses of AIDS. The median CD4+ cell count was 35 mm3. Only 29 patients are evaluable for response, since 8 received only one cycle of chemotherapy. Fifteen of 29 (52%) patients obtained objective responses, with only 4 (14%) achieving complete remissions (CR) of 1, 4, 14 and 29+ months. Three (16%) CRs were achieved in 19 evaluable patients included in the study because of poor PS, and only one CR was observed in 10 evaluable patients with histories of OI, either alone or with poor PS. The most common side effect was bone marrow toxicity with 2 related toxic deaths.(ABSTRACT TRUNCATED AT 250 WORDS)