In 1986, the bone marrow transplant centers in Ontario agreed to a strategy for the treatment of patients with NHL. Suitable patients would undergo autotransplant but be referred for allotransplant if they had persistent marrow involvement or an inadequate marrow/stem cell harvest. Data of all patients were recorded in a database. We reviewed this database to compare these transplant modalities with respect to overall survival, rate of relapse and treatment-related mortality. Between January 1986 and August 1997, 429 patients underwent BMT for NHL - 385 autotransplants and 44 allotransplants. Sixty-eight percent of patients received their transplant for aggressive NHL, while the others had indolent lymphoma. Three-year actuarial survival did not differ between allogeneic and autologous BMT: 71% vs 62%, respectively (P = 0.5330 by log-rank testing). Three-year actuarial rate of relapse was lower after allotransplant than autotransplant: 6% vs 41%, respectively (P = 0.0006 by log-rank testing). Treatment-related mortality was higher after allotransplant than autotransplant: 23% vs 6%, respectively (P = 0.001 by chi2 analysis). For further comparison, autotransplant patients were randomly matched 2:1 with the allotransplant patients for age +/- 5 years, disease status at BMT, disease histology, and year of BMT. In the matched comparison, survival did not differ (relative risk of death after allotransplant: 0.711 (95% CI: 0.309-1.637)). Relapse rate was significantly lower in the allotransplant group (relative risk of relapse for allotransplant: 0.190 (95% CI: 0.043-0.834)) and treatment-related mortality was not significantly different (relative risk for allotransplant: 1.425 (95% CI: 0.527-3.851)). In conclusion, a review of a provincial strategy for treatment of NHL, shows that survival is not different after allogeneic or autologous BMT, but the rate of relapse is lower after allotransplant. These data support continuing the current provincial strategy.