We retrospectively reviewed 71 histopathologically-confirmed bone and soft-tissue metastases of unknown origin at presentation. In order to identify the site of the primary tumour all 71 cases were examined with conventional procedures, including CT, serum tumour markers, a plain radiograph, ultrasound examination and endoscopic examinations, and 24 of the 71 cases underwent 2-deoxy-2-[F-18] fluoro-D-glucose positron emission tomography (FDG-PET). This detected multiple bone metastases in nine patients and the primary site in 12 of the 24 cases; conventional studies revealed 16 primary tumours. There was no significant difference in sensitivity between FDG-PET and conventional studies. The mean maximal standardised uptake value of the metastatic tumours was significantly higher than that of the primary tumours, which is likely to explain why FDG-PET did not provide better results. It was not superior to conventional procedures in the search for the primary site of bone and soft-tissue metastases; however, it seemed to be useful in the staging of malignancy.