BACKGROUND AND OBJECTIVE: The management of non-Hodgkin's lymphomas (NHL) in AIDS is difficult because of the poor bone marrow reserve and immunosuppression of these patients. Combination chemotherapy is the treatment of choice in the subset of patients with good performance status and mild immunosuppression. Several combination chemotherapy regimens have been used in these patients but the results have been poor. We have studied the clinical and biological features, response to treatment, outcome and prognostic factors of 46 patients with NHL and HIV infection, diagnosed in a single institution between January 1988 and June 1997. DESIGN AND METHODS: Forty-six patients with NHL and HIV infection were treated with CHOP. Patients with previously treated systemic NHL, primary CNS NHL, performance status > 2 and active AIDS-defining opportunistic or neoplastic diseases were excluded. The parameters evaluated were: age, sex, risk activity, basic hematologic and biochemical parameters, CD4 lymphocyte count, B symptoms, stage (Ann Arbor), histologic subtype (REAL classification), International Prognostic Index (IPI), response to treatment, relative dose intensity (RDI), relapse free survival (RFS) and overall survival (OS). Uni- and multivariate analyses of prognostic factors were performed. RESULTS: Median age was 35 years and 40 patients were male. CD4 lymphocyte count was lower than 0.1 x 10(9)/L in 18 out of 38 cases, hypoalbuminemia was registered in 24 (52%), serum LDH was higher than 400 U/L in 20 (43%) and beta 2-microglobulin was higher than the normal range in 9 out of 20 patients (45%). Complete response was achieved in 18 patients (40%). Twenty-six patients received G-CSF after chemotherapy. Grade 4 neutropenia and fever were significantly more frequent in patients who did not receive G-CSF. Serum LDH > 400 U/L and hypoalbuminemia were the only parameters associated with a lower probability to achieve complete response (p = 0.015 and p = 0.025, respectively). The median RFS was 26 (6-47) months and no variable was found to have statistically significant influence on it. The median OS was 9.2 (4.5-14) months, and IPI score 1 and ESR < 60 mm/h were the only parameters identified as good prognostic factors in the multivariate analysis (p = 0.03 and 0.049, respectively). INTERPRETATION AND CONCLUSIONS: In spite of patient selection, the response to CHOP treatment in patients with NHL and HIV infection remains poor. Episodes of neutropenic fever are less frequent when G-CSF is administered after CHOP. The IPI score 1 is the most important favorable prognostic factor for survival.