Using a vascularized cornea rabbit model closely resembling human high-risk keratoplasty, corneal allografts were performed on three groups of animals that were paired and tested preoperatively by mixed lymphocyte cultures (MLCs). The three groups were group A, unmodified controls with clear corneas; group B, untreated animals with vascularized corneas; and group C, animals treated with topical cyclosporine (CSA) with vascularized corneas. The MLC results were expressed as stimulation indices (SIs) and divided into low (SI < or = 20) and high (SI > 20) responders and were correlated with final outcome of grafts using survival analysis estimates. In group A, five of 13 (38.5%) grafts rejected, the chance of failure depending on the degree of MLC mismatch between donor and recipient (p = 0.02). All allografts in group B rejected regardless of the degree of mismatch. In group C, seven of 12 (58.3%) grafts rejected, indicating that topical CSA significantly improved survival (p = 0.003) compared with group B. Grafts with mild degrees of MLC mismatch (low responders) survived better (p = 0.0003) than did higher degrees of MLC mismatch (high responders), all of which rejected despite treatment. Our results indicate that both corneal vascularization and the degree of donor-recipient matching play important roles in determining corneal graft survival.