Our present study focuses on the identification of predictive biomarkers in serum for the early diagnosis of endometriosis in a minimally invasive manner using (1)H-NMR based metabonomics. PLS-DA modeling of bins obtained from CPMG spectra of serum samples discriminated endometriosis patients from controls with sensitivity and specificity levels of about 80% and 90%, respectively. Compared with those from controls, serum samples from endometriosis patients showed increased levels of lactate, 3-hydroxybutyrate, alanine, leucine, valine, threonine, lysine, glycerophosphatidylcholine, succinic acid and 2-hydroxybutyrate as well as decreased levels of lipids, glucose, isoleucine and arginine. Our work offers valuable information for non-invasive diagnosis of endometriosis and may be of potential benefit to understand pathogenesis of the disease.