A thoracoscopic examination was performed in 41 patients under local anesthesia in the lateral decubitus position. Prior thoracocentesis (38 patients) and blind biopsy with an Abrams' needle (32 patients) had been nondiagnostic. The initial nine patients were examined with the flexible fiberoptic bronchoscope, yielding a diagnostic accuracy of 56 percent (five cases). This technique was discontinued when two patients had normal findings on biopsies, despite the visual observation of later diagnosed carcinoma. Subsequent thoracoscopic procedures were performed with a rigid 11-mm single-puncture thoracoscope (Storz), which was diagnostic in 28 (88 percent) of the remaining 32 patients. A hemothorax (400 ml) was the only potentially serious complication. Twelve patients were prospectively monitored during the thoracoscopic procedure for changes in cardiac rhythm and oxygen saturation. Sinus tachycardia was the only arrhythmia observed. The mean fall in oxygen saturation was 1.4 percent. We conclude that thoracoscopic examination with the rigid thoracoscope is diagnostically superior to the fiberoptic bronchoscope and is a safe procedure which can be performed under local anesthesia.