OBJECTIVES: We examined the accuracy of the Ottawa Ankle Rule (OAR) to rule out ankle and mid-foot fractures in patients presenting with acute ankle sprain and differences of accuracy between surgeons and non-surgeons. DESIGN: Prospective cohort study. SETTING: Swiss urban secondary care centre. PARTICIPANTS: Between September 2001 and October 2002 359 patients presented with a case of ankle sprain. Of these, 251 patients both met recruitment criteria and provided data for this study. A group of surgeons and non-surgeons assessed the OAR and all patients underwent blinded radiographic assessment. MAIN OUTCOME MEASURES: Sensitivity, specificity of the OAR. RESULTS: Of the 251 patients with ankle sprains 33 had an ankle fracture (13%) and none had a mid-foot fracture. All cases with a fracture had a positive OAR result (sensitivity 100% 95% CI; 89-100) and of 218 patients without a fracture, the OAR was negative in 45 cases (specificity 21%; 16-27). In the subgroup of patients assessed by surgeons, sensitivity was 100% (77-100) and specificity was 32% (20-46). In the non-surgical group, sensitivity was also 100% (82-100) but specificity was lower (17% (11-23). CONCLUSIONS: This validation study of the OAR in a Swiss setting produced similar results than those published previously in various other settings. We found differences in the performance of the rule between surgical and non-surgical staff indicating that the OAR has its interpretation component which is more difficult to judge properly by well-instructed non-surgical assessors.