Abstract Background Surgical resection for pancreatic cancer offers the only chance of cure. Assessment of the resectability of a pancreatic tumour is therefore of great importance. The aim of the study was to investigate whether centre of diagnosis influences the likelihood of surgery and whether this affects long-term survival. Methods Patients diagnosed with non-metastasized pancreatic cancer (M0) between 2005 and 2013 in the Netherlands were selected from the Netherlands Cancer Registry. Hospitals were classified as a pancreatic centre (at least 20 resections/year) or a non-pancreatic centre (fewer than 20 resections/year). The relationship between centre of diagnosis and likelihood of surgery was analysed by multivariable logistic regression. Influence of centre on overall survival was assessed by means of multivariable Cox regression analysis. Results Some 8141 patients were diagnosed with non-metastasized pancreatic cancer, of whom 3123 (38·4 per cent) underwent surgery. Of the 2712 patients diagnosed in one of 19 pancreatic centres, 52·4 per cent had exploratory laparotomy compared with 31·4 per cent of 5429 patients diagnosed in one of 74 non-pancreatic centres (P < 0·001). A pancreatectomy was performed in 42·8 and 24·6 per cent of the patients respectively (P < 0·001). Multivariable analysis revealed that patients diagnosed in a pancreatic centre had a higher chance of undergoing surgery (odds ratio 2·21, 95 per cent c.i. 1·98 to 2·47). Centre of diagnosis was not associated with improved long-term survival (hazard ratio 0·95, 95 per cent c.i. 0·91 to 1·00). Conclusion Patients with non-metastasized pancreatic cancer had a greater likelihood of having surgical treatment when the diagnosis was established in a pancreatic centre.