Endometriosis (EMs) is a common gynecological disease accompanied by metabolic disturbances. However, the causality between metabolites and the risk of EMs remains unclear. We conducted a 2-sample Mendelian randomization (MR) analysis using the publicly available genome-wide association study (GWAS) of 486 circulating metabolites and EMs. The inverse variance weighted (IVW) was mainly used for assessing causality. MR-Egger intercept, MR-PRESSO Global, leave-one-out, and Cochran Q test analyses were used for sensitivity analyses. A total of 25 causal metabolites related to EMs have been identified, including 13 known and 12 unknown ones. Among the known metabolites, caffeine (OR = 0.86, 95% CI: 0.76-0.98, P = .026), cortisol (OR = 0.64, 95% CI: 0.41-0.99, P = .047), glycocholate (OR = 0.67, 95% CI: 0.51-0.87, P = .003), adrenate 22:4n6 (OR = 0.52, 95% CI: 0.35-0.77, P = .001), and ergothioneine (OR = 0.62, 95% CI: 0.47-0.81, P = .000) were protective factors for EMs, while mannose (OR = 1.43, 95% CI: 1.01-2.03, P = .044), 4-acetamidobutanoate (OR = 1.92, 95% CI: 1.27-2.89, P = .002), 1-linoleoylglycerol (OR = 1.36, 95% CI: 1.10-1.68, P = .005), bilirubin (Z, Z) (OR = 1.15, 95% CI: 1.01-1.31, P = .032), threonate (OR = 1.42, 95% CI: 1.14-1.77, P = .002), bilirubin (E, E) (OR = 1.18, 95% CI: 1.01-1.38, P = .039), erythronate (OR = 1.59, 95% CI: 1.01-2.52, P = .047), and dimethylarginine (SDMA + ADMA) (OR = 2.07, 95% CI: 1.19-3.62, P = .010) were risk factors for EMs. Additionally, there was no evidence of heterogeneity or pleiotropy of the known metabolites. Leave-one-out analysis indicated that the MR findings were robust. Our findings provide valuable circulating biomarkers as well as therapeutic targets for the screening, prevention, and treatment of EMs.