The purpose of this study was to evaluate fluorine-18 fluorodeoxyglucose positron emission tomography (FDG-PET) for the detection of soft tissue and bone infections. Forty-five PET examinations in 39 patients (26 male, 13 female, age range 27-86 years) with suspected infectious foci were examined with whole- or partial-body PET scans using FDG. Twenty-seven scans were done in patients with soft tissue and 18 in patients with bone infections. Corrected and uncorrected transaxial PET images were acquired. Seven hundred and twelve body regions in these 45 PET scans were evaluated. Pathological findings were graded using a confidence scale from A to E (A, definitive infection; E, no infection). Disease status was defined in all patients by culture, biopsy or surgery and clinical follow-up. In 45 PET scans there were 40 true-positive, four false-positive and one false-negative findings. Twelve foci suspected to be infectious in nature on the basis of other imaging examinations were identified as negative by PET, thus representing true-negative findings. Sensitivities for the patients with soft tissue (STI) and bone infections (BI) and for the pooled data were 96%, 100% and 98%, respectively. As the calculation of specificity is not straightforward, it was calculated on a per lesion as well as on a per body region basis to permit estimation of an upper and a lower limit. On a per lesion basis, specificities were 70% (STI), 83% (BI) and 75% for the pooled data and on a per body region basis (dividing the body into 22 regions) they were 99% (STI), 99% (BI) and 99% for the pooled data. One false-negative result was found in a patient with cholangitis. It is concluded that PET appears to be a highly sensitive method to detect infectious foci. Specificity is more difficult to estimate, but is probably in the range from 70% to above 90%.