Abstract The most frequent cause of human corneal transplant failure is immunologic rejection. Surface antigenic determinants on nucleated cells are thought to prime the host's immune system for rejection. Greater than 90% of the nucleated cells of the cornea reside in the epithelium; therefore, one might expect that host sensitization and subsequent graft rejection could be modified by epithelium removal prior to transplantation. The major objective of this two-institution combined prospective and retrospective clinical study was to determine the effect of epithelium removal on the incidence of human corneal allograft rejection reaction. A retrospective study of 152 patients showed an incidence of 24.7% rejection reactions in patients with epithelium transplanted compared to 7.2% in the epithelium-removed group (P = 0.008). In the prospective study of 55 patients, there was a rejection incidence of 30% in the epithelium-on group versus 8.0% with epithelium removed (P = 0.04). The decision to remove graft epithelium, however, must be tempered by anticipation of postoperative healing problems.