Small series have suggested that Fluorodesoxyglucose Positron-Emission-Tomography with Computed-Tomography (FDG-PET/CT) is feasible to screen for cancer in patients with unprovoked venous thromboembolism (VTE), but without validation in a large population. The aim was to assess diagnostic accuracy indices of FDG-PET/CT for occult cancer diagnosis in patients with unprovoked VTE. We analysed patients from the FDG-PET/CT group of a randomized trial that compared a screening strategy based on FDG-PET/CT with a limited screening strategy for occult malignancy detection in patients with unprovoked VTE. FDG-PET/CT was interpreted as positive for cancer, as negative or as equivocal. Patients were considered as having cancer on the basis of screening results, or of any test performed during a two-years follow-up period. We ran two sets of analysis, considering patients with equivocal FDG-PET/CT as positive, then as negative for malignancy. Between March 2009, and August 2012, 172 patients were included. FDG-PET/CT was interpreted as positive for malignancy in 10 patients (5.8%), as equivocal in 23 patients (13.4%) and as negative in 139 patients (80.8%). Malignancy was diagnosed in 7/10 (70.0%), 2/23 (8.7%) and 1/139 (0.7%) patients, respectively. Grouping positive and equivocal results, sensitivity and specificity were 90% (95%CI 60% to 98%) and 85% (95%CI 79% to 90%), respectively. Grouping negative and equivocal results, sensitivity and specificity were 70% (95%CI 40% to 89%) and 98% (95%CI 95% to 99%), respectively. FDG-PET/CT showed good accuracy for occult cancer screening in patients with unprovoked VTE. Remaining challenges include the need to define specific interpretation criteria in this dedicated population.