Abstract: Objective：To investigate the high-accuracy predictive factors of ovarian response to ovulation induction. Methods：The basic information, medication and outcome of 5 859 IVF/ICSI cycles were respectively analyzed in our center from June 2014 to December 2016. The Logistic regression model was developed to predict the ovarian response. There were a total of 517 cycles in the low-response group, 1 281 cycles in the high-response group and 4 061 cycles in the normal-response group. The clinical data of 1 143 cycles in 2017 were used to test the diagnostic value of those factors. Results：The differences in the age, AFC, basal hormone levels, medication and clinical outcomes were significant among the three groups (P＜0.001). The level of serum AMH was the important predictor of ovarian response, with the cut-off values of 2.76 ng/mL for high ovarian response and 1.50 ng/mL for low ovarian response. AMH if combining with age could predict high ovarian response, while AMH combining with AFC also predict low ovarian response. Conclusions：The level of serum AMH was a good predictor of ovarian response before Gn initiation. AMH if combining with AFC and age was of great significance in obtaining an appropriate number of oocytes, ensuring a good pregnancy outcome and preventing ovarian hyperstimulation syndrome.

Key words: Ovulation induction, Fertilization in vitro, Sperm injections, intracytoplasmic, Anti-Mullerian hormone, Ovarian response