Neoadjuvant chemotherapy (NACT) is widely used to treat esophageal squamous cell carcinoma with lymph node metastasis (ESCC). However, NACT frequently has differential effects on primary tumor (PT) and lymph node metastasis (LNM). The clinical significance of this phenomenon remains unclear. Reduction in tumor size of PT and LNM was evaluated separately in 47 node-positive ESCC patients undergoing NACT, followed by surgical resection. We analyzed the prognostic significance and various clinicopathological parameters. NACT resulted in an average reduction rate of 45.5% for PT and 36.6% for LNM; the correlation between these rates was weak but significant (r(2) = 0.122, P = 0.016). The reduction rates in both PT and LNM were significant prognostic factors, with the maximal significance with cut-off at 30% size reduction for PT (3-year survival, 47.3 vs. 8.3%, P = 0.0004) and 20% for LNM (51.3 vs. 7.1%, P = 0.0013). When these cut-off values were used to define NACT response, 28 patients (59%) were deemed responders for both PT and LNM, while 7 (15%) were nonresponders for both, and the response was inconsistent in 12 patients (26%). Only both PT/LNM responders showed good survival rates, with the remaining categories showing poor survival (3-year survival 60.5 vs. 5.3% P < 0.0001). Multivariate analysis identified neither the PT nor the LNM response alone as an independent prognostic factor; however the combined PT/LNM response was identified as an independent prognostic factor (hazard ratio [HR] 2.861, P = 0.0255) in addition to the number of histological lymph node metastases (HR 2.551, P = 0.0328). The response to NACT in LNM and PT correlates closely with postoperative survival. A good response in both enhances the postoperative prognosis.