Between 1988 and 1992, 60 patients with intermediate and high-grade non-Hodgkin's lymphomas (NHL) were treated with a new multidrug combination chemotherapy including 4'epidoxorubicin (25 mg/m2), etoposide (60 mg/m2), cyclophosphamide (400 mg/m2), administered intravenously (i.v.) on the 1st, 2nd and 3rd day every 4 weeks, prednisone (40 mg/m2) orally for 6 days every 4 weeks, vincristine (1 mg/m2) i.v. and methotrexate (400 mg/m2) i.v. on the 8th day every 4 weeks, vindesine (2.5 mg/m2) and cytarabine (200 mg/m2) on the 15th day every 4 weeks. Patients achieving apparent complete remission (CR) or good partial response (PR) after the 1st cycle of therapy were submitted to three other cycles of the same therapy. Patients failing to respond to the 1st cycle or whose disease progressed despite therapy, were treated with an alternative 2nd line therapy. Seventeen patients (28%) had stage II-II E, 15 (25%) stage III and 28 (47%) stage IV disease. Tumoral mass > 10 cm was found in 28 cases, the presence of extranodal sites (ES) in 32 cases, serum lactate dehydrogenase (LDH) > 240 IU/l in 34 cases, performance status (PS) > or = 2 in 12 cases. CR was obtained in 46 (76.4%) out of the 60 patients. Relapse-free survival (RFS) was 82, 64 and 61% with a median follow-up of 12, 24 and 36 months respectively. No relapse occurred later than 26 months after achievement with CR thus far. Overall survival (OS) was 77% at 12 months and calculated to be 62% and 59% at 24 and 36 months, respectively. Two patients died as a result of the treatment. Reversible myelosuppression was the main toxic effect. One hundred and ten out of the 221 cycles of chemotherapy were delayed because of therapy toxicity. Negative prognostic factors on the RFS and OS were the presence of an advanced stage of disease, a mass larger than 10 cm, the presence of ES, the elevated LDH, the PS > or = 2, the delay of therapy. In conclusion, results obtained using our protocol overlap those from other third generation regimens. Toxicity was also similar. The influence of clinical conditions such as stage of disease, the presence of ES, high LDH level and tumoral mass > 10 cm on the RFS and OS were significant. Principal variables influencing prognosis must be unified to compare results of similar treatments from different institutions.