It is currently unknown what degree of human leukocyte antigen (HLA)-mismatching is acceptable in unrelated donor hematopoietic stem cell transplantation (UD-HSCT). Mismatches at some loci may be more permissive than others. We have analyzed the effect of high-resolution HLA-matching on outcome of all 214 consecutive recipients of UD-HSCT carried out in Switzerland. All typing was by the Swiss reference laboratory. Donor-recipient pairs were HLA-10/10 matched (n=130) or mismatched for either HLA-A/-B/-DRB1/multiple loci (n=33; (HLA-A/-B=10); (-DRB1=8); (multiple=15)); HLA-C (n=29) or HLA-DQ/-DRB3 (n=22; (DQ=16); (-DRB1=6)). The median follow-up was 32 months. Survival probabilities (+/-95% confidence interval) at 3 years were 57 (+/-10)% for recipients of HLA 10/10-matched transplants, 53 (+/-22)% for recipients of HLA-DQ/-DRB3-mismatched transplants, 44 (+/-20)% for recipients of HLA-C-mismatched transplants and 0% for recipients of transplants mismatched at HLA-A/-B/-DRB1/multiple loci (P<0.0001). In multivariate analyses, HLA compatibility was the variable most significantly associated with survival and treatment-related mortality. We found important differences in survival in recipients of UD-HSCT with best results for transplants from 10/10 matched donors. Single mismatches at HLA-DQ/-DRB3 were well tolerated, mismatches at HLA-C had intermediate results and mismatches at HLA-A/-B/-DRB1/multiple loci resulted in poor survival.