A study was undertaken to evaluate the safety and efficacy of thoracoscopic lung biopsy for interstitial lung disease. The relation between operative findings, pathologic findings, and preoperative computed tomographic scan findings was examined. Twenty-six patients, 10 male and 16 female, underwent thoracoscopic lung resection to diagnose interstitial lung disease. Sixteen patients were outpatients for an elective procedure; 10 were inpatients including 2 who were ventilator dependent. The mean length of operation was 54 minutes and the mean length of chest tube duration, 1.3 days. There were no deaths. Staphylococcal pneumonia developed in 1 patient postoperatively. One patient with systemic pulmonary hypertension was ventilator dependent for 48 hours. A double-lumen endotracheal tube was used in all but 2 patients. Twelve-millimeter trocar ports were used to allow easy interchange of staplers and endoscopic instruments. Biopsy of at least two lobes was performed in each patient with resection of a piece of grossly abnormal lung. A single chest tube was left routinely. The pathologic diagnosis was usual interstitial pneumonitis in 7 patients. Four patients had interstitial fibrosis and 4, granulomas. Three patients had diffuse alveolar damage and 3, Wegener's granulomatosis. Two patients had bronchiolitis obliterans with organizing pneumonia. One patient each had lymphangioleiomyomatosis, eosinophilic granuloma, and cytomegalovirus. Sixteen patients underwent preoperative computed tomographic scanning. The scans were assessed by 2 radiologists who were blinded to the surgical results. Computed tomography accurately predicted the site of disease in most instances. Four patients had at least one lobe with no evidence of disease on computed tomography but with interstitial lung disease found thoracoscopy.(ABSTRACT TRUNCATED AT 250 WORDS)