We conducted a prospective study to update our knowledge of fever of unknown origin (FUO) and to explore the utility of a structured diagnostic protocol. From December 2003 to July 2005, 73 patients with FUO were recruited from 1 university hospital (n = 40) and 5 community hospitals (n = 33) in the same region in The Netherlands. FUO was defined as a febrile illness of >3 weeks' duration, a temperature of >38.3 degrees C on several occasions, without a diagnosis after standardized history-taking, physical examination, and certain obligatory investigations. Immunocompromised patients were excluded. A structured diagnostic protocol was used. Patients from the university hospital were characterized by more secondary referrals and a higher percentage of periodic fever than those referred to community hospitals. Infection was the cause in 16%, a neoplasm in 7%, noninfectious inflammatory diseases in 22%, miscellaneous causes in 4%, and in 51%, the cause of fever was not found (no differences between university and community hospitals). There were no differences regarding the number and type of investigations between university and community hospitals. Significant predictors for reaching a diagnosis included continuous fever; fever present for <180 days; elevated erythrocyte sedimentation rate, C-reactive protein, or lactate dehydrogenase; leukopenia; thrombocytosis; abnormal chest computed tomography (CT); and abnormal F-fluorodeoxyglucose positron emission tomography (FDG-PET). For future FUO studies, inclusion of outpatients and the use of a set of obligated investigations instead of a time-related criterion are recommended. Except for tests from the obligatory part of our protocol and cryoglobulins in an early stage, followed by FDG-PET, and in a later stage by abdominal and chest CT, temporal artery biopsy in patients aged 55 years or older, and possibly bone marrow biopsy, other tests should not be used as screening investigations.