OBJECTIVE: We aimed to investigate the optimal cutoff value of circumferential resection margin (CRM) of esophageal squamous cell carcinoma (ESCC) in patients who underwent radical esophagectomy. BACKGROUND: Tumor involvement of a CRM in ESCC has not been clearly defined. METHODS: We reviewed 479 pT3 ESCC patients to find the optimal cutoff point of distance from CRM in addition to 0 μm for discriminating survival time. RESULTS: The partitions at and near the 500 μm distance from CRM generated the largest log-rank statistics (P = 0.0086). Therefore, we added 500 μm as an additional cutoff value for a positive CRM. Compared to patients with CRM greater than 500 μm, patients with CRM 0 μm showed worse overall survival (P < 0.001) and progression-free survival (P < 0.001), followed by patients with 0 to 500 μm (P = 0.008 and 0.066, respectively). In multivariable analyses, overall survival differences remained significant [0 < CRM ≤ 500 μm vs CRM > 500 μm, hazards ratio (HR) = 1.875, 97.5% CI: 1.243-2.829, P = 0.002; CRM = 0 μm vs CRM > 500 μm, HR = 2.666, 97.5% CI: 1.745-4.076, P < 0.001]. In comparison of criteria from the College of American Pathologists, the Royal College of Pathologists, and this study, HRs of positive CRM (95% CI, P-value) were 1.969 (1.501-2.584, P < 0.001), 1.384 (1.039-1.844, P = 0.027), and 1.696 (1.342-2.143, P < 0.001), respectively. CONCLUSIONS: In patients with ESCC, we developed new, 3-tiered CRM criteria providing more detailed prognostic information than the 2-tiered criteria.