To determine the usefulness of [18F]fluorodeoxyglucose (FDG) whole body FDG-PET in the diagnosis of tumours in patients with paraneoplastic neurological syndromes (PNS), we prospectively studied 20 patients with paraneoplastic antibodies in whom conventional imaging gave negative or inconclusive results for the presence of tumour. All 20 patients had neurological manifestations compatible with PNS and well-characterized paraneoplastic antibodies (12 anti-Hu, one anti-Hu and anti-CV2, one anti-CV2, four anti-Yo, one anti-Ri and one anti-amphiphysin). The mean delay between the onset of neurological symptoms and FDG-PET was 10 months (range 1-54). In these 20 patients, abnormal uptake was demonstrated in 18 patients, with some patients having abnormal signal in several areas. We observed abnormal uptake in the mediastinum (13 cases), lung (two cases), breast (two cases), parotid gland (one case), or the cervical, supraclavicular or axillary lymph nodes (seven cases). Following FDG-PET, the histological diagnosis of the tumour was made in 14 patients (small cell lung carcinoma in eight cases, breast adenocarcinoma in two, lung adenocarcinoma in two, axillary metastasis of ovary carcinoma in one, and malignant thymoma in one). Two other patients with abnormal FDG uptake showed radiological evidence of lung cancer, but a histological diagnosis could not be obtained. In two other patients, initial FDG-PET showed abnormal FDG uptake that was not confirmed a few months later by repeat FDG-PET. In the two patients with negative FDG-PET, peritoneal carcinomatosis was diagnosed in one and no tumour was found in the other. In our series, the sensitivity of FDG-PET for tumour detection was >83% demonstrating a clear role of this technique in the management of patients with PNS. However, in our series, the specificity of FDG uptake was only 25% due to unexplained abnormal FDG uptake in three patients and in abnormal FDG uptake due to a benign tumour in one patient. Over the study period, we saw 73 other patients with PNS and paraneoplastic antibodies. A tumour was demonstrated in 71 out of 73 by conventional techniques. Since false-positive and false-negative results are possible with FDG-PET and in most patients with PNS, the tumour is demonstrated by conventional techniques, we believe that FDG-PET should be reserved, at the moment, for patients with well-defined PNS antibodies when conventional imaging fails to identify a tumour or when lesions are difficult to biopsy.