OBJECTIVE Mediastinal lymph node staging determines the treatment strategy for non-small-cell lung cancer. This study aims to evaluate prospectively the accuracy of preoperative integrated 18-fluoro-2-deoxy-D-glucose positron emission tomography-computerised tomography ((18)FDG PET-CT) for mediastinal lymph node staging. METHODS Preoperative integrated (18)FDG PET-CT was used to analyse mediastinal lymph nodes in patients with non-small-cell lung cancer. Nodal stations were identified according to the American Thoracic Society mapping system. Lymph nodes with a standardised uptake value (SUVmax) >3 were considered to be positive. The mediastinal lymph nodes were harvested during lung resection and the results of integrated (18)FDG PET-CT were compared to the mediastinal lymph node histology results. RESULTS A total of 51 patients were enrolled in this study. The mean interval between integrated (18)FDG PET-CT and surgery was 31+/-15.8 days (range: 2-78 days). The mean mediastinal lymph node harvested and station number per patient during surgery were 11.8+/-5.6 (range: 2-27) and 3.8+/-1 (range: 2-6), respectively. The incidence of N2 pathological disease was 19.6%. The integrated (18)FDG PET-CT sensitivity and specificity were 40+/-30% and 85+/-11%, respectively. The positive and negative predictive values were 40+/-30% and 85+/-11%, respectively. False-positive results (six patients) were mainly due to inflammatory lymph nodes. False-negative results (six patients) were mainly due to infra-centimetrical, malignant lymph node invasion. CONCLUSIONS The sensitivity of integrated (18)FDG PET-CT for mediastinal lymph node staging in patients selected for surgery is low. When positive mediastinal lymph nodes are detected, invasive mediastinal staging must be performed. On the other hand, the specificity is high: patients with negative integrated (18)FDG PET-CT can be operated upon without invasive mediastinal staging.