The purpose of the study was to compare conventional cyclophosphamide/doxorubicin/vincristine/prednisolone (CHOP) chemotherapy with CHOP (3 courses) plus etoposide/methylprednisolone/high-dose cytarabine/cisplatin (ESHAP), high-dose therapy (HDT), and autologous peripheral blood progenitor cell transplantation (PBPCT) as front-line treatment for poor-prognosis aggressive non-Hodgkin's lymphoma (NHL). Between May 1, 1995, and April 30, 1998, 58 patients, aged 15-55 years, newly diagnosed with poor-prognosis aggressive NHL (category F-H by the Working Formulation) were enrolled. According to the age-adjusted international prognostic index, 65% of the patients were high-risk cases and 35% made up the high-intermediate group. After 3 courses of CHOP, 25 of 48 patients were randomized to continue with CHOP, and 23 were randomized to receive 2-4 cycles of ESHAP followed by HDT and PBPCT. There was no significant difference in the rate of complete remission between the two groups (36%, 95% confidence interval [CI]: 18%-57% in CHOP vs. 43%, 95% CI: 23%-65% in ESHAP/HDT) (P = 0.77). With a median follow-up duration of 39 months, the 4-year failure-free survival (FFS) was superior in the ESHAP/HDT group (38%, 95% CI: 18%-58% vs. 15%, 95% CI: 4%-32%) (P = 0.04). The disease-free survival was marginally different in favor of the ESHAP/HDT arm (90%, 95% CI: 47%-98% vs. 37%, 95% CI: 7%-69%) (P = 0.06). The 4-year overall survival between the two treatment arms was comparable (51%, 95% CI: 28%-70% for ESHAP/HDT vs. 30%, 95% CI: 13%-48% for CHOP) (P = 0.25). Treatment-related mortalities were not significantly different between both groups (17%, 95% CI: 5%-39% for ESHAP/HDT vs. 8%, 95% CI: 1%-26% for CHOP) (P = 0.41). However, only 61% of the patients assigned to the ESHAP/HDT arm underwent HDT and PBPCT. As compared with CHOP, the corporate regimen of CHOP/ESHAP/HDT seems to improve the FFS in patients with newly diagnosed, poor-prognosis aggressive NHL.