Abstract Background The aim was to define the pathological response in lymph nodes following neoadjuvant chemotherapy for oesophageal adenocarcinoma and to quantify any associated survival benefit. Methods Lymph nodes retrieved at oesophagectomy were examined retrospectively by two pathologists for evidence of a response to chemotherapy. Patients were classified as lymph node-negative (either negative nodes with no evidence of previous tumour involvement or negative with evidence of complete regression) or positive (allocated a lymph node regression score based on the proportion of fibrosis to residual tumour). Lymph node responders (score 1, complete response; 2, less than 10 per cent remaining tumour; 3, 10–50 per cent remaining tumour) and non-responders (score 4, more than 50 per cent viable tumour; 5, no response) were compared in survival analyses using Kaplan–Meier and Cox regression analysis. Results Among 377 patients, 256 had neoadjuvant chemotherapy. Overall, 68 of 256 patients (26·6 per cent) had a lymph node response and 115 (44·9 per cent) did not. The remaining 73 patients (28·5 per cent) had negative lymph nodes with no evidence of regression. Some patients had a lymph node response in the absence of a response in the primary tumour (27 of 99, 27 per cent). Lymph node responders had a significant survival benefit (P < 0·001), even when stratified by patients with or without a response in the primary tumour. On multivariable analysis, lymph node responders had decreased overall (hazard ratio 0·53, 95 per cent c.i. 0·36 to 0·78) and disease-specific (HR 0·42, 0·27 to 0·66) mortality, and experienced reduced local and systemic recurrence. Conclusion Lymph node regression is a strong prognostic factor and may be more important than response in the primary tumour.