Background: Upper gastrointestinal malignancies have a poor prognosis. There is no consensus on how patients should be followed after surgery. The authors hypothesized that a structured follow-up programme including endoscopic ultrasonography (EUS) and [18F]fluorodeoxyglucose (FDG) PET/CT would detect cancer recurrences, leading to more patients being eligible for therapy. Methods: After surgery with curative intent for adenocarcinomas in the gastro-oesophageal junction, stomach or pancreas, patients were randomized 1 : 1 to standard clinical assessment in the outpatient clinic at 3, 6, 9, 12, 18 and 24 months after operation, or clinical assessment plus imaging including [18F]FDG PET/CT and EUS. The primary endpoint was number of patients receiving oncological treatment for recurrence. Secondary endpoints were overall and progression-free survival, survival after recurrence detection of isolated locoregional recurrences and risk factors affecting survival. Results: In total, 183 patients were enrolled, including 93 who underwent standard follow-up and 90 who had follow-up plus imaging. A recurrence was detected in 84 patients within 2 years after surgery (42 in each group), including 33 of 42 patients in the imaging group who were asymptomatic. Some 25 of 42 patients in the imaging group and 14 of 42 in the standard group received chemotherapy (P = 0·028). Although survival after detection of recurrence in asymptomatic patients was significantly longer than that for symptomatic patients (P < 0·001), overall survival from date of surgery in the two treatment groups was comparable. Conclusion: Follow-up after surgery for upper gastrointestinal cancer with EUS and PET/CT leads to detection of more asymptomatic cancer recurrences and patients referred for treatment without prolonging overall survival. Registration number: NCT02209415 (http://www.clinicaltrials.gov).