Microsatellite analyses show that self-reported ethnicity often correlates poorly with true genetic ancestry. As unknown ancestral differences could potentially have an impact on transplant outcome, we developed an average allele length discrepancy (AALD) score to assess allele length discrepancy between donor/recipient (D/R) using microsatellites analysed routinely in post-transplant chimeric assessment. This was then compared with outcome in a homogeneously treated cohort of pediatric patients undergoing high-resolution sibling or matched unrelated donor transplantation for acute lymphoblastic leukemia (ALL). AALD scores formed a numeric continuum ranging from 0 to 1.4 (median 0.76) for sibling pairs and 0.8-2.17 (median 1.6) for high-resolution matched unrelated donor (HR-MUD) pairs. There was a trend for worse OS with increasing AALD score, which reached statistical significance above a threshold of 1.7 for OS. Patients whose transplants had an AALD score of ⩾1.8 had a risk of non-relapse mortality 4.9 times greater (P=0.025) and relapse risk three times greater (P=0.058) than those scoring <1.8. This approach will now be explored in a Centre International for Blood and Marrow Transplantation Research (CIBMTR) study of 750 D/R pairs across all disease groups; if confirmed, it has the potential to improve donor selection for patients with multiple prospective donors.