Twenty-five years ago over 90% of children with acute lymphoblastic leukaemia (ALL) died of this disease. Dramatic improvement has been achieved since then by employing risk-adapted, aggressive polychemotherapy protocols. More than 90% of children with ALL treated according to, for example BFM-protocols, have nowadays cure rates in the range of 70%-80%. However, 10% of patients do not initially respond adequately to standard induction chemotherapy. They are characterized by distinct chromosomal abnormalities such as translocation (9; 22) or combinations of early treatment failure and other risk factors as cytogenetic abnormalities, lineage-specific surface markers or tumour load at diagnosis. In this group of patients in first complete remission and certainly in the vast majority of relapsed patients, allogeneic bone marrow transplantation (BMT) has evolved as an alternative approach allowing further intensification of myeloablation and the introduction of an additional antileukaemic alloreactivity. Nevertheless, the decision for a marrow transplant in children has to be made very carefully because of a significant increase in treatment related mortality and BMT-specific risks like acute and chronic graft-versus-host disease with a critical iatrogenic chronic morbidity. This is even more evident, if mismatched or unrelated transplants are being considered. The indications for one or the other treatment modality according to the current BFM strategy are discussed.