Patients with aggressive non-Hodgkin's lymphoma who fail to achieve a complete remission (CR) with standard induction therapy have a poor prognosis with conventional-dose salvage therapy alone. Retrospective series have suggested that early introduction of high-dose salvage therapy with autologous stem cell transplantation (ASCT) may benefit partial-responder (PR) patients. However, two randomized studies (of 69 and 51 patients with partial clinical responses) failed to demonstrate any advantage of intensive therapy. By contrast, the GELA comparative study on 94 PR-patients (residual disease being histologically documented in 53 patients) suggested that high-dose therapy with ASCT improves survival. Interpretation of all these results is complicated by the heterogeneity of patient populations with respect to initial prognostic factors, induction regimens and, in particular, the criteria used to define partial response. Gallium CT scan and magnetic resonance imaging are now used to better explore residual masses. In the future, early restaging with these imaging techniques might be used to delineate patients who are likely to achieve CR from those who will fail to induction treatment and could be candidates for experimental treatments.