This cohort study investigated whether the risk of attrition during Australian Army recruit training was predicted by the fitness, age, date of enlistment, or injury status of recruits. Subjects were 1,317 male Australian Army recruits undertaking 12 weeks of intensive training. Fitness was measured using a 20-m progressive shuttle run test (20 mSRT) in which higher scores reflected higher fitness. A total of 184 subjects failed to complete training. Two hundred seventy-six disabling lower limb training injuries were recorded; 100 were stress fractures or periostitis. Scores on the 20 mSRT ranged between 3.5 and 13.5. Multivariate survival analysis revealed a strong negative association between 20 mSRT score and risk of attrition (p < 0.001) and a positive association between sustaining a lower limb injury and risk of attrition (p < 0.001). These effects were additive. Age and enlistment date were not significantly associated with risk of attrition. Fitness and training procedures may be important, modifiable risk factors for attrition.