No consensus has been achieved regarding the optimal extent of lymph node (LN) dissection for node-negative ESCC patients. This study aimed to determine the optimal extent of LN dissection for node-negative ESCC patients. We retrospectively reviewed 481 ESCC patients with node-negative resection and no preoperative therapy. Overall survival (OS) was evaluated by the log-rank test and multivariate Cox regression. The 5-year OS was 51.7% and 64.7% for patients with 1-5 and ≥6 negative LNs resected, respectively (P<0.001). However, there was no significant survival difference between patients with 6-12 negative LNs resected and patients with over 12 negative LNs resected (P=0.205). Multivariate analysis indicated that the negative LN count was independently associated with better survival. In the subgroup analysis, no optimum lymphadenectomy was defined for T1; the minimum number of LNs that needed to be resected was 6 nodes for T2 and 7 nodes for T3. No survival benefit was observed when extensive lymphadenectomy was performed. The nomogram, including the number of LNs examined, T stage, and histologic differentiation, had more predictive power than TNM staging. The results of our study suggest that ESCC patients with LN-negative tumors should have at least 6 LNs examined for T2 and 7 LNs for T3, but extensive lymphadenectomy is not recommended. The nomogram, including the number of LNs examined, T stage, and histologic differentiation, is a useful clinical tool.