<p><b>BACKGROUND</b>Thoracoscopy is highly sensitive and accurate for detecting pleural effusions. However, most respiratory physicians are not familiar with the use of the more common rigid thoracoscope or the flexible bronchoscope, which is difficult to manipulate within the pleural cavity. The semi-rigid thoracoscope combines the best features of the flexible and rigid instruments. Since the practice with this instrument is limited in China, the diagnostic utility of semi-rigid thoracoscopy (namely medical thoracoscopy) under local anesthesia for undiagnosed exudative pleural effusions was evaluated.</p><p><b>METHODS</b>In 50 patients with undiagnosed pleural effusions who were studied retrospectively, 23 received routine examinations between July 2004 and June 2005 and the rest 27 patients underwent medical thoracoscopy during July 2005 and June 2006. Routine examinations of the pleural effusions involved biochemistry and cytology, sputum cytology, and thoracentesis. The difference in diagnostic sensitivity, costs related to pleural fluid examination and complications were compared directly between the two groups.</p><p><b>RESULTS</b>Medical thoracoscopy revealed tuberculous pleurisy in 6 patients, adenocarcinoma in 7, squamous-cell carcinoma in 2, metastatic carcinoma in 3, mesothelioma in 2, non-Hodgkin's lymphoma in 1, and others in 4. Only 2 patients could not get definite diagnoses. Diagnostic efficiency of medical thoracoscopy was 93% (25/27). Only 21% patients were diagnosed after routine examinations, including parapneumonic effusion in 2 patients, lung cancer in 2 and undetermined metastatic malignancy in 1. Twelve patients with tuberculous pleurisy were suspected by routine examination. Costs related to pleural effusion testing showed no difference between the two groups (P=0.114). Twenty-three patients in the routine examination group underwent 97 times of thoracentesis. Two pleural infection patients and 2 pneumothorax patients were identified and received antibiotic treatment and drainage. Medical thoracoscopy could be well tolerated by all the patients. The semi-rigid thoracoscope could be easily controlled by chest physicians. The most common complication was transient chest pain (20 of 27 patients) from the indwelling chest tube, which would be managed with conventional analgesics. One case of subcutaneous emphysema and 2 cases of postoperative fever were self-limiting. No severe complications occurred.</p><p><b>CONCLUSIONS</b>Medical thoracoscopy is a simple, safe, and cost-effective tool, with a high positive rate. Physicians should extend its access to proper patients if the facilities for medical thoracoscopy are available.</p>