The use of high-dose chemotherapy with or without total-body irradiation (TBI) followed by autologous hematopoietic cell transplantation is associated with improved survival for relapsed or refractory non-Hodgkin's lymphoma (NHL). Previous reports comparing preparatory regimens with or without TBI followed by autologous bone marrow transplantation (ABMT) or peripheral blood progenitor cell transplantation (PBPCT) for these patients did not demonstrate any survival difference between the different modalities. No randomized studies comparing survival for patients with NHL transplanted with radiochemotherapy vs. chemotherapy alone have been reported. We treated 221 patients with high-risk, relapsed or refractory NHL with either chemotherapy alone or radiochemotherapy followed by ABMT or PBPCT. The patients were assigned preparatory regimens in a non-randomized manner and this analysis was performed to evaluate differences in outcome with the two preparatory regimens. Actuarial five-year event-free survival (EFS) was similar in patients receiving fractionated total-body irradiation (FTBI) plus etoposide (VP-16) and cyclophosphamide (Cy) compared with chemotherapy alone consisting of carmustine (BCNU) plus identical doses of VP-16 and Cy (52% vs. 46%, p = 0.08). Overall survival (OS) favored radiochemotherapy (61%) compared with chemotherapy alone (53%, p = 0.02). The relapse rate was the same in both groups (41%), whereas the transplantation-related mortality (TRM) was similar in patients receiving chemotherapy alone and those receiving radiochemotherapy (13% vs. 7% respectively, p = 0.30). Proportional hazards analysis of significant variables including preparatory regimen found only the number of prior relapses to be predictive of EFS. Fewer number of prior relapses, radiochemotherapy and PBPCT were significant predictors of favorable OS. In additional analyses, the improved OS of the radiochemotherapy regimen was confirmed only for patients receiving ABMT but was not a significant predictor of outcome in patients transplanted with PBPCT. From these retrospective data we conclude: 1) PBPCT resulted in survival superior to that of ABMT; 2) the risk of relapse is similar with either preparatory regimen; 3) patients with fewer prior relapses enjoyed superior overall and event-free survival as well as fewer relapses; and 4) there were no significant differences in the two preparatory regimens when combined with PBPCT in relapsed or refractory NHL.