Objective: The current AJCC staging system for esophageal cancer is based on lymph node location, irrespective of the number of involved and examined lymph nodes. Here we review our experience with N stage subclassification.Methods: We enrolled 488 patients (94% with squamous cell carcinoma) receiving primary curative resection without neoadjuvant therapy for esophageal cancer between 1995 and 2006. The impact of total resected lymph node number (TLN), metastatic lymph node number (MLN) and ratio (MLR) on patient survival was investigated.Results: The overall 3-year survival rate was 35.4% for the entire population. The 3-year survival rate was equivalent among patients in N1 (23.3%), M1a (22.0%), and nonregional lymph node metastasis-related M1b (18.5%, p = 0.321). There was no survival difference between patients with TLN <15 and TLN ≥15 (Figure 1A, p = 0.249). In contrast, both MLN and MLR significantly predicted patient survival. The 3-year survival rate was 52.3%, 29.2%, and 8.0% for patients with MLN = 0, 1-3, and ≥4, respectively (Figure 1B, p < 0.001). For patients with MLR = 0-0.2 or >0.2, the 3-year survival rate was 28.7% and 9.8%, respectively (Figure 1C, p < 0.001). However, survival rate differences became insignificant when TLN was less than 15 (MLN = 1-3 vs. ≥4: p = 0.252; MLR = 0-0.2 vs. >0.2: p = 0.109). Conclusions: We recommend designating both regional and nonregional lymph nodes as N nodes and reserving the M1 stage for distant organ metastasis. Examinaton of a minimum of 15 lymph nodes is required for nodal staging subclassification. MLN and MLR, but not TLN, predict survival in esophageal cancer.