OBJECTIVE: To investigate the value of positive lymph node ratio (LNR) in predicting the prognosis of patients with esophageal cancer. METHODS: We retrieved the data of a total of 862 patients with esophageal cancer with complete clinical pathology data archived in SEER database in 2010 to 2015. The best cutoff point of LNR was selected using X-tile software. Univariate and multivariate COX proportional hazard models were used to assess the value of LNR in predicting the prognosis of patients after propensity score matching (PSM). RESULTS: The best cut-off point of LNR determined using X-tile 3.6.1 software was 0.16. The patients with LNR < 0.16 and those with LNR≥0.16 showed significant differences in the number of positive lymph nodes, pathological type, T stage and M stage. After 1:1 propensity score matching, the two groups showed no significant difference in the clinical data or pathological parameters. Matched univariate and multivariate COX regression analyses showed that LNR, primary tumor site and M staging were all independent risk factors affecting the prognosis of patients, and among them LNR had the most significant predictive value (LNR < 0.16 LNR≥0.16: HR=1.827, 95% : 1.140-2.929; =0.000). The median survival time of patients with LNR < 0.16 was 31 months (95%: 22.556-39.444 months), as compared with 16 months (95%: 12.989-19.011) in patient with LNR≥0.16 (Log Rank χ=27.392, < 0.0001). LNR had a better accuracy than N stage for assessing the patients' prognosis with an area under the ROC curve of 0.617 (95%: 0.567-0.666), as compared with 0.515 (95%: 0.463-0.565) of N stage (=3.008, =0.0026). CONCLUSIONS: LNR≥0.16 is an independent risk factor affecting the prognosis of patients with esophageal cancer and has better prognostic value than N stage.