High-dose therapy with autologous stem cell transplantation (ASCT) has been widely proposed for patients with relapsed Hodgkin's disease (HD). From 1982 to 1993, we selected (from the French registry for bone marrow transplantation) 280 patients, who underwent ASCT for relapsed HD after initial treatment including chemotherapy. Patient characteristics at diagnosis were: sex ratio (M/F): 1.5; median age: 30 years (5-59 years), stage I, II: 43%; III, IV: 57%; 32% had chemotherapy, 68% chemo+ radiotherapy. All patients achieved complete remission after first-line therapy and subsequently relapsed. The median interval between diagnosis and high-dose therapy was 34 months. First relapse occurred in 78% of the patients at a median end-of-treatment to relapse time of 18 months. All patients received salvage chemotherapy before high-dose therapy, and the median time between relapse and high-dose therapy was 5 months. After this regimen, 84% of the patients were considered to have chemosensitive relapse. Conditioning regimens were: BEAM: 60%; CBV/BEAC: 26%. Transplant-related mortality was 6%. With a median follow-up of 3 years after high-dose therapy, overall and progression-free survivals at 4 years were 66 and 60%, respectively. Neither the conditioning regimen nor the stem cell source affected survival. Good prognostic factors for survival were: chemosensitivity of relapse (P < 0.01) and first relapse vs further relapse (P < 0.05). For 214 patients in first relapse, other significant factors for survival were: end-of-treatment to relapse interval < 12 months (P < 0.05) and nodal vs extranodal relapse (P < 0.001). These two prognostic factors were used to validate a prognostic model with three significantly different subgroups: 0 (n = 59), 1 (n = 125), or 2 factors (n = 30) with 4-year survival, respectively, at 93, 59 and 43% (P < 0.001). Salvage therapy can be tailored in patients with relapsing HD: conventional treatment in the good prognosis group (0 factor), high-dose therapy after response to second line regimen (1 factor) and more intensive treatments for the bad prognosis group (2 factors).