Relapsed acute lymphoblastic leukaemia (ALL) is the most common cause for a fatal outcome in paediatric oncology. Although initial ALL cure rates have improved up to 80%, the prognosis of recurrent ALL remains dismal with event-free-survival (EFS) rates about 35%. In order to analyse a population-based cohort with uniform treatment of initial disease, we examined the outcome of children suffering from relapsed ALL in Austria for the past 20 years and the validity of the currently used prognostic factors (e.g. time to and site of relapse, immunophenotype). Furthermore, we compared survival rates after chemotherapy alone with those after allogeneic stem cell transplantation (SCT). All 896 patients who suffered from ALL in Austria between 1981 and 1999 were registered in a prospectively designed database and treated according to trials ALL-Berlin-Frankfurt-Münster (BFM)-Austria (A) 81, ALL-A 84 and ALL-BFM-A 86, 90 and 95. Of these, 203 (23%) suffered from recurrent disease. One-hundred-and-seventy-two patients (85%) achieved second complete remission. The probability of 10-year EFS for the total group was 34 +/- 3%. Clinical prognostic markers that independently influenced survival were time to relapse, site of relapse and the immunophenotype. Additionally, a Cox regression model demonstrated that allogeneic SCT after first relapse was associated with a superior EFS compared with chemo/radiotherapy only (hazard ratio = 0.254; P = 0.0017).