A randomized study was carried out in 36 patients to determine the diagnostic value of abdominal computerized tomography (CT) in protracted fever of obscure origin and/or unexplained inflammatory syndromes. There were no clinical or biological signs pointing to a diagnosis, and 55% of the patients had already been unsuccessfully explored in other hospital departments. CT was performed initially in group I patients and after some of 56 predetermined conventional examinations (including 20 considered "invasive") carried out in a definite order in group II patients. An early diagnosis could be made within 2 months in 17 patients (47%): 7 in group E and 10 in group II. The underlying disease was diagnosed in 4 cases in group I and 1 case in group II (where 4 other cases could have been diagnosed, had CT been performed earlier); the difference was not significant. Nor was there any difference between the two groups in the number of examinations performed, the cost of investigation and the duration of stay in hospital. However, a number of reasons are in favour of early CT examination in protracted fever: the so-called "invasive" explorations were uncomplicated in group I and complicated in 2 patients of group II; following randomization, a diagnosis was made by CT within less than 3 days in the 4 patients of group I, and there were neither false-positive nor false-negative results of CT. If CT had been performed early in all patients, 9 diagnoses would have been made, representing 25% of the cases or 1/2 of early diagnoses. Abdominal CT proved unable to establish the 10 late diagnoses (i.e. after a mean delay of 18 months).