Approximately 20% of pleural effusions remain without an established aetiology after evaluation. Thoracoscopy has a very high sensitivity for the diagnosis of both benign and malignant diseases and greatly increases the diagnostic yield for pleural effusion. We sought to evaluate the diagnostic yield and safety of medical thoracoscopy at this institution. The records of all patients undergoing medical thoracoscopy for the evaluation of undiagnosed pleural effusion between 1990 and 1996 were reviewed. The procedure was performed under local anaesthesia with sedation using a Stortz rigid thoracoscope. Fifty-eight patients had thoracoscopy, most having had two (range: 1-6) non-diagnostic pleural aspirations and biopsies of the pleura. Nineteen patients were found to have mesothelioma and nine metastatic malignancy. Three patients were considered likely to have tuberculous pleural disease, six had asbestos related benign pleural fibrosis and three post-cardiotomy syndrome. There was one chylous effusion of uncertain aetiology, one posttraumatic and two other benign effusions, both of which resolved without clear aetiology. On seven occasions the pleural space could not be adequately accessed, but none of these patients had prior computerized tomography (CT) or ultrasound of the pleural space. There were five false negative diagnoses of malignancy, but no false positives. The diagnostic sensitivity for pleural malignancy was 85% and specificity 100%. There were no major complications, but four patients had late tumour seeding at the thoracoscopy site. Medical thoracoscopy is a safe procedure with a high diagnostic yield. Pre-operative evaluation of the pleural collection using ultrasound or CT increases the likelihood of successful access to the pleural space and may increase diagnostic yield.