To analyse the risk factors of relapse before bone marrow transplantation (BMT) and to present the prognostic information as good as possible.Methods A total of 3142 patients, who underwent the allogeneic blood or bone marrow tran sp lantation between 1989 and 1997 and were documented in the European Group for Bl ood and Marrow transplantation (EBMT), were included. Six possible risk factors including type of donor, stage of disease, age, gender, donor? recipient sex co mbination and the waiting time from diagnosis to transplation of relapse were co nsidered. The time to relapse was analysed by Kaplan Meier curves and Cox regre ssion with stratification on prognostic factors that did not satisfy the Proport ional Hazard Assumption.Results An amount of 447 patients relapsed out of all 3142 patients. The relapse rate wa s 14.2%. Type of donor and stage of disease showed a clear prognostic effect, bu t failed the proportional hazard assumption. Therefore, the data were stratified on the combination of type of donor and stage of disease. Within these strata a n additional significant effect of age could be observed. Relative risk of age ≥40 vs age <40 was 1.32 (95% confidence interval 1.09-1.59). The prognostic model is summarized graphically.Conclusions The combination of type of donor, stage of disease and age of recipient at tr ansplantation are important prognostic factors for relapse after BMT. To analyze the risk factors of relapse before bone marrow transplantation (BMT) and to present the prognostic information as good as possible. Methods A total of 3142 patients, who underwent the allogeneic blood or bone marrow tran splantation between 1989 and 1997 and were recorded in the European Group for Blod and Marrow transplantation (EBMT), were included. Six possible risk factors including type of donor, stage of disease, age, gender, donor? co-recipient sex co mbination and the waiting time from diagnosis to transplation of relapse were co nsidered. The time to relapse was analysed by Kaplan Meier curves and Cox regre ssion with stratification on prognostic factors that did not satisfy the Proportional Hazard Assumption. Results An amount of 447 patients relapsed out of all 3142 patients. The relapse rate wa s 14.2%. Type of donor and stage of disease showed a clear prognostic effect, bu t failed the proportional hazard assumption. Therefore, the data were stratified On the combination of type of donor and stage of disease. Within these strata an additional significant effect of age could be observed. Relative risk of age ≥ 40 vs age <40 was 1.32 (95% confidence interval 1.09-1.59). The prognostic model is summarized graphically. Conclusions The combination of type of donor, stage of disease and age of recipient at tr ansplantation are important prognostic factors for relapse after BMT.