We analysed data for 213 patients with ALL and AML who received either peripheral blood stem cells (PBSC) (n=74) or bone marrow (BM) (n=139) from an HLA-matched unrelated donor (EBMT acute leukaemia registry; January 1994 to January 1999). The two groups of patients (by cell source) were comparable with respect to age, sex, disease status, year at transplant and graft T cell depletion. Engraftment was achieved in about 90% regardless of stem cell source or leukaemia type. Kinetics of neutrophil and platelet recovery, similar for both sources in ALL patients, were faster for PBSC in AML patients. The incidence of acute graft-versus-host disease was similar for both sources in AML patients, but higher for PBSC in ALL patients (74 vs 54%, P=0.05). The 1-year probability of chronic graft-versus-host disease was 40 and 45% (P=0.66) in ALL patients compared to 49 and 35% (P=0.13) in AML patients (PBSC vs BM). In AML patients, none of the following differed significantly with cell source: transplant-related mortality, relapse incidence, leukaemia-free survival and overall survival. In ALL patients, the transplant-related mortality for PBSC vs BM was 61 vs 47% (P=0.13), the relapse incidence was 47 vs 39% (P=0.17), the leukaemia-free survival was 21 vs 32% (P=0.04) and the overall survival was 24 vs 34% (P=0.04). These data suggest that the short-term outcome of allogeneic PBSC is not significantly different from that of BM in AML patients who underwent a transplant from a matched unrelated donor but, conversely, that survival with PBSC may be decreased in ALL patients. In conclusion, the source of transplant cells needs to be evaluated by disease, especially when dealing with unrelated donors.