Pain is a common unspecific symptom in orthopaedic prosthetics. The accurate differentiation between synovitis, loosening or infection is often difficult with conventional X-rays, arthrography or bone scintigraphy. Because of the high glucose uptake of inflammatory cells, [18F]fluorodeoxyglucose (18F-FDG) is an appropriate tracer for the evaluation of suspected inflammation or infection. In this preliminary study we describe 18F-FDG PET findings in patients referred for evaluation of painful hip or knee prostheses. We studied 23 patients with 28 prostheses, 14 hip and 14 knee prostheses, who had a complete operative or clinical follow-up. 18F-FDG PET scans were obtained with an ECAT EXACT HR+ PET scanner. High glucose uptake in the bone prostheses interface was considered as positive for infection, an intermediate uptake as suspect for loosening, and uptake only in the synovia was considered as synovitis. The imaging results were compared with operative findings or clinical outcome. PET correctly identified three hip and one knee prostheses as infected, two hip and two knee prostheses as loosening, four hip and nine knee prostheses as synovitis, and two hip and one knee prostheses as unsuspected for loosening or infection. In three patients covered with an expander after explantation of an infected prosthesis PET revealed no further evidence of infection in concordance with the clinical follow-up. PET was false negative for loosening in one case. Our preliminary results suggest that FDG PET could be a useful tool for differentiating between infected and loose orthopaedic prostheses as well as for detecting only inflammatory tissue such as synovitis.