Outcome of allogeneic BMT for childhood ALL performed at a national centre was evaluated for the period between 1985 and 1996. Sixty-seven patients representing all Danish children and adolescents (1-19 years) transplanted for ALL, were evaluated. Patients transplanted with a family donor (n = 51) had a 3-year probability of leukaemia-free survival (P-LFS) of 56% (95% confidence limits 41-69%) and patients receiving marrow from a matched unrelated donor (MUD) (n = 16) had a 3-year P-LFS of 67% (34-86%) (P = 0.38). Relapse was responsible for 48% of the deaths and occurred with increasing frequency among children transplanted with marrow from HLA-identical siblings. Patients transplanted with family donors from 1985-1989 had P-LFS of 72% (54-84%) compared to patients transplanted after 1990 who revealed a significantly reduced P-LFS of 28% (10-49%, P = 0.01). Furthermore, the risk of relapse was increased (65% (24-88%)) (P = 0.02) in the later period. In contrast, patients transplanted with marrow from a MUD (1990-1996) had a lower risk of relapse (11%, 5-20%) (P = 0.002) but a comparable risk of transplant-related mortality. Children who experience relapse after modern, intensive treatment for ALL may have an increased propensity for relapse even after BMT. Therefore, when performing BMT for childhood ALL, there may be a benefit in overall survival from the increased graft-versus-leukaemia effect that follows less intensive GVHD prophylaxis or transplantation with a MUD.