Objective: To analyze the different metabolites and related metabolic pathways in follicular fluid and embryo culture fluid of polycystic ovary syndrome (PCOS) and non-PCOS groups. Finding markers predictable for clinical outcomes of in vitro fertilization-embryo transfer (IVF-ET) treatment. Design: Single center cross-sectional study Setting: Center for Reproductive Medicine, Zhengzhou University First Affiliated Hospital. From September 2019 to October 2019 Population and sample: 60 women who underwent IVF-ET were selected, including 30 with PCOS and 30 with the fallopian tubal issues only. We collected the first tube of follicular fluid (FF) and the wasted embryo culture medium (ECM) of all women. Methods: All samples were performed nontargeted Ultra High Performance Liquid Chromatography-Mass Spectrometry (UHPLC-QE-MS) analysis. Related metabolic pathways were screened by KEGG annotation. To search potential indicators, the logistic regression was made combined with clinical data. Mean outcome measures: Predictive performance of markers for clinical outcomes (pregnancy rate, delivery rate, live birth rate, miscarriage rate). Results: Comparing the PCOS group against the non-PCOS group, we found 11 significantly different metabolites in the FF and 56 in the ECM. And a total of 11 kinds of biomarkers associated with clinical outcomes. Androsterone sulfate, Glycerophosphocholine, and Elaidic carnitine seem robust to predict the abortion rate of the PCOS group, with AUC of 0.941, 0.933, 0.933, respectively. The glycerol phospholipid metabolic pathway is enriched in both the follicular fluid and embryo culture fluid. Conclusions: The differential metabolites were mainly a variety of lipids. Some of them can predict clinical outcomes to a certain extent.