AimEnhanced recovery programmes (ERPs) involve early postoperative oral feeding. The aim of this study was to test the hypothesis that intolerance to early feeding was associated with a complicated postoperative course.MethodA retrospective cohort analysis of the prospective multicentre database developed by the Francophone Group for Enhanced Recovery after Surgery (GRACE) was undertaken. Seventy‐one centres in Belgium, France and Switzerland participated in the study. All patients were encouraged to eat within 24 h after surgery. Patients were separated into two groups according to whether early feeding was well tolerated (WT) or poorly tolerated (PT). The primary outcome measure was overall postoperative complications. Secondary outcome measures were unplanned reoperation, early mobilization rate and duration of postoperative hospital stay.ResultsAmong the cohort of 3034 patients, early feeding was WT in 2614 patients (WT group) and PT in 420 patients (PT group). There were significantly more postoperative complications in the PT group than in the WT group (52.1% vs 17.0%, respectively; P = 0.001), namely more unplanned reoperations, less early mobilization and longer postoperative hospital stay. Multivariate analyses confirmed that PT early feeding was the main and dominant independent factor for postoperative complications [OR 4.47 (95% CI3.49–5.72); P < 0.001], more unplanned reoperations and longer hospital stay.ConclusionsThis study demonstrates a close relationship between intolerance to early feeding and a complicated postoperative course. Whenever this simple very early red flag is observed, discharge should not be planned until postoperative complications have been ruled out.