OBJECTIVES There is uncertainty about patient selection and the adequate timing at which fluorine-18 fluorodeoxyglucose (F-FDG) PET/computed tomography (CT) is indicated in the diagnostic work-up of fever of unknown origin (FUO). The aim of this study was to determine the diagnostic performance of F-FDG PET/CT in patients with FUO. METHODS All consecutive patients who underwent F-FDG PET/CT at the University Hospital Zurich because of FUO between 2006 and 2012 were included in this retrospective, observational study. RESULTS A total of 76 patients [70% men, median (interquartile range) age 60 (47-67) years] were included. F-FDG PET/CT showed characteristically increased F-FDG activity in 56 patients (74%), leading to confirmation of or change in the suspected cause of FUO in 57 and 17%, respectively. The final diagnosis after F-FDG PET/CT included infection (21%), malignancy (22%), noninfectious inflammatory disease (12%), others (5%), or an unknown cause (40%). The success rate, sensitivity, and specificity of F-FDG PET/CT were 60, 77, and 31%, respectively. Sensitivity was highest in patients with suspected malignancy (100%, 95% confidence interval 79-100%). Diagnostic performance was independent of the investigated variables other than suspected infection as a cause of FUO (odds ratio 0.1, 95% confidence interval 0.01-0.8, P=0.033). CONCLUSION The diagnostic performance of F-FDG PET/CT was significantly higher in patients with suspected malignancy causing a FUO compared with suspected infection or noninfectious inflammatory disease. However, it was independent of the baseline characteristics and duration of fever. This supports the recommendation to perform F-FDG PET/CT early in the diagnostic work-up of FUO, which may shorten disease duration and lower health costs, particularly when infection or malignancy is suspected.