Although aggressive NHL in relapse after remission can still be cured by second-line treatment followed by high-dose therapy and autologous stem cell transplantation, the long-term prognosis of patients who fail to obtain remission after first-line therapy remains extremely poor. We retrospectively evaluated a series of 29 consecutive patients with primary refractory high-grade NHL who were treated with Dexa-BEAM (DB) as uniform salvage therapy at a single institution. Twenty-nine patients with aggressive NHL primary refractory to CHOP or CHOP-like induction therapy with a median age of 47 (range, 22 - 64) years received 1 - 2 cycles of DB and were candidates for subsequent autologous stem cell (PBSC) mobilization and transplantation (PBSCT). Follow-up of all patients was updated in March 2004. Eight of 29 patients (28%) responded to one cycle of DB (1 complete/7 partial remissions); 2 of whom are alive after PBSCT (1 autologous/1 matched unrelated donor), 1 patient died after autologous PBSCT. Reasons for failure to proceed to high-dose therapy in spite of response to DB were recurrent progressive disease (n = 2), septicemia (n = 1), and allogeneic transplant-related mortality after mobilization failure to DB (n = 2). Twenty-one patients failed to respond to DB and died of progressive disease. Overall survival was 7% after 41 months. We conclude that Dexa-BEAM salvage therapy is not effective in patients with truly primary refractory high-grade NHL. The efficiency of rituximab combined with Dexa-BEAM or novel chemotherapeutic strategies needs to be established.