Introduction: Hyper-response is a primary risk factor of ovarian hyperstimulation syndrome (OHSS), a serious side-effect of controlled ovarian stimulation (COS). Prediction of hyper-response is vital for choosing the appropriate COS protocol to avoid OHSS in these patients. Objectives: To compare the accuracy of AMH, FSH, and AFC and to indentify their cut-offs for predicting hyper-response. Methods: A prospective cohort study was conducted on 820 patients. Each patient had AMH, FSH and AFC measured on cycle day 2. Hyper-response was diagnosed when patients had 15 oocytes retrieved. Results: The rate of hyper-response was 21,8 percent. The AMH test was the most accurate in predicting hyperresponse, followed by AFC, and the least accurate predictive test was the FSH. AMH 3.57 ng/ml predicted hyper-response with a sensitivity of 83.7 percent and specificity of 79.8 percent. Conclusion: AMH has the highest accuracy in predicting hyper-response.