The effect of HLA matching in corneal transplantation is still--after numerous of studies--disputable. We investigated the effect of DRB1 matching in high-risk cases with vascularization and/or retransplantation. Only class II antigens were matched because we were unable to obtain donor lymphocytes for HLA typing. Typing was performed on DNA isolated from the ocular tissues up to 24 hr after death. When this study was initiated, DNA-based methods had been developed only for class II typing. The first part of the study concerns 74 cases with at least 3 years of observation fully matched for 17 DRB1 specificities detected using restriction fragment-length polymorphism. This showed an improved long-term graft survival of 72% compared with 45% in a historical control group of 23 comparable cases. In the second part of the study, stored DNA samples from the restriction fragment-length polymorphism-matched donor-recipient pairs were subjected to retyping with a new method based on sequence-specific polymerase chain reaction. It was possible to split DRB1*01, *04, and *11 in 3, 14, and 5 alleles, respectively. The matching was then re-assigned taking all splits into account. This showed that 36 cases had at least one incompatibility, whereas 38 cases were fully compatible. The long-term graft survival rate was 79% in the matched group compared with only 59% in the mismatched group, which is significantly different at P=0.032. This retrospective, but blinded, randomized study is strong evidence for the effect of matching and may give scope for international collaboration to obtain completely matched corneas for this group of patients.