We assessed the potential benefits of including systematic 18fluorodeoxyglucose positron emission tomography (FDG-PET) for detecting tumour recurrence in a prospective randomised trial. Patients (N=130) who had undergone curative therapy were randomised to undergo either conventional (Con) or FDG-PET procedures during follow-up. The two groups were matched at baseline. Recurrence was confirmed histologically. 'Intention-to-treat' analysis revealed a recurrence in 46 patients (25 in the FDG-PET group, and 21 in the Con group; P=0.50), whereas per protocol analysis revealed a recurrence in 44 out of 125 patients (23 and 21, respectively; P=0.60). In another three cases, PET revealed unexpected tumours (one gastric GIST, two primary pulmonary cancers). Three false-positive cases of FDG-PET led to no beneficial procedures (two laparoscopies and one liver MRI that were normal). We failed to identify peritoneal carcinomatosis in two of the patients undergoing FDG-PET. The overall time in detecting a recurrence from the baseline was not significantly different in the two groups. However, recurrences were detected after a shorter time (12.1 vs 15.4 months; P=0.01) in the PET group, in which recurrences were also more frequently (10 vs two patients) cured by surgery (R0). Regular FDG-PET monitoring in the follow up of colorectal cancer patients may permit the earlier detection of recurrence, and influence therapy strategies.