Polycystic ovary syndrome (PCOS) is a complex endocrine disorder with well-established metabolic abnormalities. In the present study, untargeted metabolomics technology was applied to analyze the serum and follicular fluid samples from women with polycystic ovary syndrome and healthy controls using 1H nuclear magnetic resonance (NMR). Seventy samples for PCOS analysis were collected in hospital of Shandong University of Traditional Chinese Medicine (Jinan, China), NMR was used as analytical technology and multivariate analysis was applied to analyze metabolomics difference in PCOS and healthy controls. Significant metabolic differences were found in both serum and follicular fluid samples with orthogonal partial least-squares discriminant analysis (OPLS-DA). Three discriminated metabolites (1-Methylhistidine, threonine and Citrate) in both serum and follicular fluid were altered in PCOS patients. Abnormal energy metabolism, lipid metabolism and amino acid metabolism were detected in PCOS patients. Furthermore, more significantly changed amino acids were discovered in follicular fluid samples. Our findings would provide a resource for further investigations on metabolic disturbance in PCOS patients.