INTRODUCTION: Anti-Mullerian Hormone (AMH) was recently introduced as a marker of ovarian reserve in assisted reproduction. The cutoff values of AMH for prediction of poor response have not yet been determined.MATERIAL AND METHODS: Ninety women undergoing their first IVF/ICSI cycle were prospectively included in this clinical, non-interventional study. Baseline AMH, follicle stimulating hormone (FSH) and antral follicle count (AFC) were measured before starting ovarian stimulation. AMH was also measured on day 5 of stimulation and in the follicular fluid of the first aspirated follicle. The predictive value of baseline AMH, day 5 AMH and follicular fluid AMH were assessed comparatively to FSH and AFC for ovarian response. Ovarian response was defined as poor (<4 oocytes), high (>12 oocytes) or normal (≥4 oocytes and ≤12 oocytes). However, only 3 patients met the criterion for high ovarian response and thus analysis was focused on the prediction of poor response.RESULTS: Significant differences were present between poor responders and non-poor responders regarding FSH (p = 0.019), baseline AMH (p = 0.002), AFC (p < 0.001), day 5 AMH (p = 0.005) but not for follicular AMH (p = 0.183). The largest AUC (area under the curve) for poor ovarian response was obtained by AFC (AUC = 0.81) followed by baseline AMH (AUC = 0.70). At a level below 2.74 ng/mL, the sensitivity of the test is 69% and specificity is 70.5%.CONCLUSION: Baseline AMH is almost as good a predictor for poor ovarian response as AFC.