[en] BACKGROUND: This study evaluates the accuracy of [F]fluorodeoxyglucose positron emission tomography (F-FDG PET) imaging with semi-quantitative analysis for differentiating benign from malignant pleural exudates and for guiding the search for the primary tumour of pleural metastases. METHODS: Whole-body 18F-FDG PET was performed in 79 patients with exudative pleurisy. Standard uptake values were normalized for body weight, body surface area, lean body mass (SUVbw, SUVbsa, SUVlbm) with and without correction for blood glucose levels. Thoracoscopy was systematically performed to reveal pathological diagnosis. RESULTS: All SUVs were significantly higher in all malignant pleural diseases (n = 51) than in benign (n = 28) (P < 0.001). Moreover SUVs were greater in the pleural metastases from pulmonary primaries (n = 25) and in mesotheliomas (n = 8) than in extrathoracic primaries (n = 18) (P < 0.01) with no significant difference between lung cancers and mesotheliomas. Receiver operating curve (ROC) analysis between benign and malignant lesions showed areas under the curves that ranged from 0.803 (SUVbsa g) to 0.863 (SUVbw). The cut-off value for SUVbw which gave the best accuracy (82.3%) was 2.2. When comparing thoracic with extrathoracic primaries the highest accuracy (80.4%) was found for a cut-off value of 2.6. CONCLUSION: Semi-quantitative analysis of 18F-FDG PET imaging helps to differentiate malignant from benign pleural exudates and to distinguish between thoracic or extrathoracic primaries.