We retrospectively studied the results from thoracenteses and needle biopsies of the pleura performed in 414 patients with pleural effusions between 1973 and 1982. The final causes of effusion were malignant disease in 281 patients (67.9%) and nonmalignant disease in 133 (32.1%). The presence of pleural malignant disease was established by cytologic study in 162 patients (57.6%), by needle biopsy in 123 (43%), and by either cytologic analysis or biopsy in 182 (64.7%). In only 7.1% of the 281 patients with malignant pleural effusions did biopsy reveal malignant disease when the results of cytologic study were negative for malignant disease. Nearly half of the patients with lymphoma had lymphocytosis of the pleural fluid, but neither this finding nor the lymphocytic pleuritis noted on biopsy was diagnostic of lymphomatous involvement of the pleura. Among the patients with malignant mesothelioma, thoracotomy was necessary to confirm the diagnosis in 60.9%. In the patients with nonmalignant diseases, with the exception of six with tuberculous pleurisy, pleural biopsy was nondiagnostic even though the causes of pleural effusion were apparent from the clinical features. The causes of pleural effusion remained unknown in 82 patients (19.8%). Pleural biopsy failed to provide adequate tissue in 55 patients (13.3%). This study shows that cytologic analysis has a higher sensitivity (P less than 0.001) than needle biopsy for diagnosing malignant pleural effusions. The value of needle biopsy is limited in establishing the cause of pleural effusion that results from either malignant or nonmalignant disease, with the exception of tuberculous pleurisy.(ABSTRACT TRUNCATED AT 250 WORDS)