Eye ruptures are among the most devastating eye injuries and can occur in automobile crashes, sporting impacts, and military events, where blunt projectile impacts to the eye can be encountered. The purpose of this study was to develop injury risk functions for globe rupture of both human and porcine eyes from blunt projectile impacts. This study was completed in two parts by combining published eye experiments with new test data. In the first part, data from 57 eye impact tests that were reported in the literature were analyzed. Projectile characteristics such as mass, cross-sectional area, and velocity, as well as injury outcome were noted for all tests. Data were sorted by species type and areas were identified where a paucity of data existed, based on the kinetic and normalized energy of assaulting objects. For the second part, a total of 126 projectile tests were performed on human and porcine eyes. Projectiles used for these tests included blunt aluminum projectiles, BBs, foam pellets, Airsoft pellets, and paintballs. Data for each projectile were recorded prior to testing and high-speed video was used to determine projectile velocity prior to striking the eye. In part three the data were pooled for a total of 183 eye impact tests, 83 human and 100 porcine, and were analyzed to develop the injury risk criteria. Binary logistic regression was used to develop injury risk functions based on kinetic and normalized energy. Probit analysis was used to estimate confidence intervals for the injury risk functions. Porcine eyes were found to be significantly stronger than human eyes in resisting globe rupture (p=0.01). For porcine eyes a 50% risk of globe rupture was found to be 71,145 J/m², with a confidence interval of 63,245 J/m² to 80,390 J/m². Human eyes were found to have a 50% risk of globe rupture at a lower, 35,519 J/m², with confidence intervals of 32,018 J/m² to 40,641 J/m². The results presented in this paper are useful in estimating the risk of globe rupture when projectile parameters are known, or can be used to validate computational eye models.