In 1989 we carried out a trial comparing allogeneic BMT to chemotherapy (CT) in 76 children with relapsed acute lymphoblastic leukaemia (ALL). Ten years on we have clinically revised outcome to firmly establish the role of each treatment, to analyse the importance of length of first remission and to provide long-term actuarial results for disease-free survival (DFS) and relapse rate in each group. For 21 patients within the transplantation group, probability of DFS and relapse are 42.8 +/- 10.8% and 40.2 +/- 11.7% (s.e.), respectively. In the chemotherapy group, probability of DFS is 10.0 +/- 4.74% (P = 0.001) and probability of relapse 87.5 +/- 5.2% (P = 0.0004). These results strongly reflect those at initial analysis, confirming a key role of BMT in the management of ALL in second remission. Moreover, on univariate analysis only two factors influenced DFS: treatment group and length of first complete remission (less or more than 30 months from first CR). Thus, it seems clear that the best therapeutic option in early relapse is BMT, whereas DFS in late relapse is at the limit of significance (P = 0.07), with a higher relapse rate in the CT group. Although encouraging results using intensified rotational combination chemotherapy have been published, prospective randomised studies are needed to assess with certainty the best therapeutic option in these patients.