Three hundred fifty male recruits were divided into two groups: Walk (N = 170) and Run (N = 180). A physical screen was performed before the commencement of recruit training consisting of a visual assessment of the feet looking for pes planus, pes cavus, and deformities of the toes. Each recruit was also asked if he had sustained any injury in the preceding 2 years. The visual findings and history of prior injury were noted and compared with actual injuries sustained during the 12-week training course. Fifty-three subjects in the Walk group and 54 in the Run group were identified as being at risk as a consequence of the screen. The sensitivity of the screen for predicting the subsequent injury was 34.4% in the Walk group and 31.8% in the Run group. The specificity was 72.6 and 77.4% in the Walk and Run groups, respectively. The predictive value of the test was 44.9% in the Walk group and 50.9% in the Run group. When reinjury was examined, the sensitivities fell to 9.4% (Walk) and 4.7% (Run) and the specificities to 57.5% (Walk) and 50.5% (Run). The screen correctly identified only 1 of 10 medical discharges in the Walk group and 2 of 16 in the Run group. The screening examination had poor sensitivity, specificity, and predictive value, and more than half of those thought to be at risk did not subsequently sustain an injury. Anecdotal beliefs that improvements in medical screening would reduce recruit wastage were not borne out. Abnormalities of the foot (pes planus, pes cavus, hallux valgus) were not significant factors in the development of injury during recruit training.