Background In sports involving pivoting and landing, female athletes suffer knee injury at a greater rate than male athletes. Hypotheses Proprioceptive deficits in control of the body's core may affect dynamic stability of the knee. Female, but not male, athletes who suffered a knee injury during a 3-year follow-up period would demonstrate decreased core proprioception at baseline testing as compared with uninjured athletes. Study Design Cohort study (prognosis); Level of evidence, 2. Methods Study subjects were 277 collegiate athletes (140 female, 137 male) who were prospectively tested for core proprioception by active and passive proprioceptive repositioning. Athletes were monitored for injury for 3 years. An ANOVA and multivariate logistic regression were used to test whether core proprioception was related to knee injuries in athletes. Results Twenty-five athletes sustained knee injuries (11 women, 14 men). Deficits in active proprioceptive repositioning were observed in women with knee injuries (2.2°) and ligament/meniscal injuries (2.4°) compared with uninjured women (1.5°, P ≤ .05). There were no differences in average active proprioceptive repositioning error between injured men and uninjured men (P ≥ .05). Uninjured women demonstrated significantly less average error in active proprioceptive repositioning than uninjured men (1.5° vs 1.7°, P ≤ .05). For each degree increase in average active proprioceptive repositioning error, a 2.9-fold increase in the odds ratio of knee injury was observed, and a 3.3-fold increase in odds ratio of ligament/meniscal injury was observed (P ≤ .01). Active proprioceptive repositioning predicted knee injury status with 90% sensitivity and 56% specificity in female athletes. Conclusions Impaired core proprioception, measured by active proprioceptive repositioning of the trunk, predicted knee injury risk in female, but not male, athletes.