The present study was designed to investigate the prognostic significance of the preoperative sensitive-modified Glasgow prognostic score (S-mGPS) and its superiority in esophageal squamous cell carcinoma (ESCC). Clinicopathologic characteristics, preoperative albumin and C-reactive protein (CRP) levels were retrospectively collected in 442 patients who underwent transthoracic esophagectomy. The S-mGPS was calculated before surgery based on optimal cutoff values of 45.6 g/L for albumin and 10.0 mg/L for CRP. 360, 74 and 8 cases were assigned an mGPS of 0, 1 and 2, respectively. In contrast, the S-mGPS was 0 in 114, 1 in 258 and 2 in 70 patients. Of the 360 patients with an mGPS of 0, 246 migrated to the S-mGPS-1 group. Both mGPS and S-mGPS were significantly correlated with tumor length, depth of invasion, pathological tumor-node-metastasis (pTNM) stage and adjuvant treatment. In addition, they were significantly associated with disease free survival (DFS) and overall survival (OS) in univariate analysis. Furthermore, multivariate Cox regression analysis identified S-mGPS as an independent prognostic indicator for both DFS [hazard ratio (HR), 1.577; 95% confidence interval (CI), 1.149-2.163; P = 0.005] and OS (HR, 1.762; 95% CI, 1.250-2.484; P = 0.001), but not mGPS (HR, 0.957; 95% CI, 0.692-1.323; P = 0.790 for DFS and HR, 1.089; 95% CI, 0.781-1.517; P = 0.615 for OS, respectively). Moreover, subgroup analysis revealed that the prognostic impact of the S-mGPS was especially striking in pTNM stage II patients. The preoperative S-mGPS is superior to the mGPS as a prognostic predictor in patients with resectable ESCC.