Background: The objective was to evaluate the impact of concentration of surgery for oesophageal and gastric cardia cancer on long-term survival in the population-based Eindhoven Cancer Registry area. In contrast to most previous studies, this study aimed to evaluate both surgically and non-surgically treated patients, to avoid the confounding effect of selective referral. Methods: This retrospective cohort study included all patients diagnosed with oesophageal or gastric cardia cancer between 1995 and 2006. Results for the period 1995-1998 were compared with those for 1999-2006, after concentration of surgery. Results: Between 1995 and 2006, 2212 patients were registered with the diagnosis, of whom 638 underwent resection. Before 1999, 73.4 per cent of surgically treated patients underwent a resection in a low-volume hospital (fewer than 4 resections per year) and 23.2 per cent were referred to an academic hospital. After concentration, 63.2 per cent of surgically treated patients underwent resection in one of two regional high-volume centres (15-20 resections per year) and 13.8 per cent were referred to an academic hospital. Three-year survival rates increased from 32.0 to 45.1 per cent for patients who had surgery (P = 0.004), and from 13.1 to 17.9 per cent for all included patients (P = 0.026). These improvements remained after adjustment for case mix or (neo) adjuvant treatments, and were similar for patients with squamous cell carcinoma or adenocarcinoma. However, adjustment for annual hospital volume attenuated this association for patients who had surgery. Conclusion: Concentration of oesophageal and gastric cardia cancer surgery was associated with improvements in long-term, population-based overall survival for surgically as well as non-surgically treated patients, apparently mediated by an increase in volume.