Abstract Background Many developments in medicine are likely to have influenced the treatment of gastrointestinal cancer, including rates of resection. This study sought to investigate changes in surgical resection rates over time among patients with gastrointestinal cancer. Methods Patients diagnosed between 1995 and 2012 in the Eindhoven Cancer Registry area were included. Multivariable logistic regression analysis was used to determine the independent influence of interval of diagnosis on the likelihood of having a resection. Results Among 43 370 patients, crude resection rates decreased between 1995 and 2012 for gastric, colonic and rectal cancer, most notably for patients aged at least 85 years with gastric cancer (from 37·3 to 13·3 per cent), and patients aged 75–84 years and 85 years or more with rectal cancer (from 80·5 to 64·4 per cent, and from 58·9 to 36·0 per cent respectively). After adjustment for patient and tumour characteristics, patients diagnosed between 2008 and 2012 with gastric (odds ratio (OR) 0·71, 95 per cent c.i. 0·55 to 0·92), colonic (OR 0·52, 0·44 to 0·62), rectal (OR 0·39, 0·33 to 0·48) and periampullary (OR 0·42, 0·27 to 0·66) cancers were less likely to undergo resection than those diagnosed between 1995 and 1998. Patients diagnosed with pancreatic cancer were more likely to undergo resection in recent periods (OR 4·13, 2·57 to 6·64). Conclusion Resection rates have fallen over time for several gastrointestinal cancers. This might reflect increased availability of other treatments, better selection of patients as a result of improved diagnostic accuracy, risk-avoiding behaviour and transparency related to surgical outcomes at hospital and surgeon level.