Reduced-intensity regimens (RIRs) are being used with increasing frequency in patients with non-Hodgkin's lymphoma (NHL) undergoing allogeneic transplantation. The impact of dose reduction on relapse and survival has not been extensively studied. We performed a retrospective analysis of 88 patients conditioned with conventional myeloablative regimens (CMRs) (n = 48) and an RIR (n = 40) of fludarabine 125 mg/m(2) and melphalan 140 mg/m(2). Compared with the patients receiving CMR, those receiving RIR were older, had more often failed autologous transplantation, and had more frequently received peripheral blood and unrelated donor transplants. Graft-versus-host disease prophylaxis was provided with cyclosporine + methotrexate +/- prednisone for the CMR and with cyclosporine + mycophenolate +/- methotrexate for the RIR. The relapse rate was significantly lower in the patients receiving CMR than in those receiving RIR (13% vs 28%; P = .05). The 1-year transplantation-related mortality rate was 33% for CMR and 28% for RIR (P = .40). Kaplan-Meier 2-year overall survival and progression-free survival were 52% and 46% for CMR versus 53% and 40% for RIR (P = not significant). Using cumulative incidence functions based on competing risks, univariate analysis, and treatment-related prognostic factors, we found that higher treatment intensity (P = .03; relative risk [RR] = 35%) and absence of previous autologous transplantation (P = .0007; RR = 20%) were associated with a lower relapse rate. Using a Cox univariate proportional hazards model, we found that chemosensitive disease at transplantation (P = .05; RR = 57%) and absence of previous autologous transplantation (P = .002; RR = 37%) were associated with improved survival. Our observation of similar survival in the patients receiving CMR and those receiving RIR confirms that RIRs are feasible alternatives for high-risk patients with NHL; however, the data suggest that reduced treatment intensity and previous autologous transplantation are associated with increased relapse.