The aim of this study was to assess the relationship between radiation therapy (RT) and treatment-related mortality in patients receiving high-dose chemotherapy (HDCT) and autologous bone marrow transplantation (ABMT) for recurrent/refractory Hodgkin's disease (HD). Between December 1986 and December 1992, 59 patients previously treated at the Princess Margaret Hospital underwent HDCT (etoposide 60 mg/kg, melphalan 160 mg/m2) and ABMT, performed for refractory (13 patients) or relapsed (46 patients) HD. RT was incorporated in the salvage treatment with the intent to achieve complete control of disease prior to ABMT. RT was given before ABMT in 33 patients, and after ABMT in 4 patients. Treatment-related (TR) mortality was defined as any death occurring within 100 days of ABMT. Autopsies were performed for all patients with TR deaths. With a median follow-up of 4.6 years (range 1.2-7.4 years), the actuarial overall survival was 41% +/- 14% at 5 years. We observed 37 deaths, and 10 of these were TR deaths. Among the 24 patients who received thoracic RT before ABMT, there were 8 TR deaths, 3 of these solely attributable to radiation pneumonitis. The remaining 5 TR deaths all had respiratory failure with complicating sepsis as a major medical problem. The interval from RT to ABMT was shorter for 8 patients dying of TR death (mean 37 days; range 0-103 days), than for the 16 survivors (mean 105 days; range 0-263 days) (P = 0.026). Among 9 patients with ABMT within 50 days of thoracic RT, 6 had TR death. In contrast, among the 35 patients without thoracic RT (26 no RT, 9 non-thoracic RT), there were only 2 TR deaths. The 4 patients treated with mantle RT post-ABMT had no serious pulmonary complications. The use of thoracic RT before HDCT and ABMT was associated with a high post-transplant mortality rate. It was most evident in patients who received thoracic RT within 50 days prior to ABMT, or when the target volume included large volume of lung. We recommend that the use of post-transplant RT be investigated to decrease TR mortality.