The use of computational modeling to predict injury mechanisms and severity has recently been investigated, but few models report failure level ligament strains. The hypothesis of the study was that models built off neutral ankle experimental studies would generate the highest ligament strain at failure in the anterior deltoid ligament, comprised of the anterior tibiotalar ligament (ATiTL) and tibionavicular ligament (TiNL). For models built off everted ankle experimental studies the highest strain at failure would be developed in the anterior tibiofibular ligament (ATiFL). An additional objective of the study was to show that in these computational models ligament strain would be lower when modeling a partial versus complete ligament rupture experiment. To simulate a prior cadaver study in which six pairs of cadaver ankles underwent external rotation until gross failure, six specimen-specific models were built based on computed tomography (CT) scans from each specimen. The models were initially positioned with 20 deg dorsiflexion and either everted 20 deg or maintained at neutral to simulate the cadaver experiments. Then each model underwent dynamic external rotation up to the maximum angle at failure in the experiments, at which point the peak strains in the ligaments were calculated. Neutral ankle models predicted the average of highest strain in the ATiTL (29.1 ± 5.3%), correlating with the medial ankle sprains in the neutral cadaver experiments. Everted ankle models predicted the average of highest strain in the ATiFL (31.2 ± 4.3%) correlating with the high ankle sprains documented in everted experiments. Strains predicted for ligaments that suffered gross injuries were significantly higher than the strains in ligaments suffering only a partial tear. The correlation between strain and ligament damage demonstrates the potential for modeling to provide important information for the study of injury mechanisms and for aiding in treatment procedure.