Abstract Background Centralization of surgery has been shown to improve outcomes for oesophageal and pancreatic cancer, and has been implemented for gastric cancer since 2012 in the Netherlands. This study evaluated the impact of centralizing gastric cancer surgery on outcomes for all patients with gastric cancer. Methods Patients diagnosed with non-cardia gastric adenocarcinoma in the intervals 2009–2011 and 2013–2015 were selected from the Netherlands Cancer Registry. Clinicopathological data, treatment characteristics and mortality were assessed for the periods before (2009–2011) and after (2013–2015) centralization. Cox regression analyses were used to assess differences in overall survival between these intervals. Results A total of 7204 patients were included. Resection rates increased slightly from 37·6 per cent before to 39·6 per cent after centralization (P = 0·023). Before centralization, 50·1 per cent of surgically treated patients underwent gastrectomy in hospitals that performed fewer than ten procedures annually, compared with 9·2 per cent after centralization. Patients who had gastrectomy in the second interval were younger and more often underwent total gastrectomy (29·3 per cent before versus 41·2 per cent after centralization). Thirty-day postoperative mortality rates dropped from 6·5 to 4·1 per cent (P = 0·004), and 90-day mortality rates decreased from 10·6 to 7·2 per cent (P = 0·002). Two-year overall survival rates increased from 55·4 to 58·5 per cent among patients who had gastrectomy (P = 0·031) and from 27·1 to 29·6 per cent for all patients (P = 0·003). Improvements remained after adjustment for case mix; however, adjustment for hospital volume attenuated this association for surgically treated patients. Conclusion Centralization of gastric cancer surgery was associated with reduced postoperative mortality and improved survival.