A prospective study of 129 consecutive thoracocentesis in 86 patients at a university medical center evaluated the clinical value, complications, and patient experience with thoracocentesis. Pleural fluid analysis in conjunction with the clinical presentation placed 78 pleural fluids into diagnostic categories: definitive 14 (18 percent), presumptive 44 (56 percent), and nondiagnostic 20 (26 percent). Fourteen of 78 (18 percent) of the nondiagnostic fluids were useful, while only six (8 percent) were not useful clinically; therefore, 92 percent of thoracocentesis provided clinically useful information. Using sequential data analysis, initial diagnostic categorizations of eight of 78 patients were upgraded from presumptive or nondiagnostic to definitive based on data available 24 hours following thoracocentesis. Thus, 70 patients were categorized based on the pleural fluid data obtained within the first 24 hours of thoracocentesis. Thirty-four objective complications occurred in 26 of 129 (20 percent) thoracocentesis. The most common complications were pneumothorax, 15 of 129 (12 percent), and cough, 12 of 129 (9 percent). Sixty-five subjective complications occurred in 56 of 123 (46 percent) thoracocentesis. Anxiety, 26 of 123 (21 percent), and site pain, 24 of 123 (20 percent), were the most common subjective complications noted. Thirty technical problems occurred in 129 (23 percent) thoracocentesis with blood contamination, 14 of 129 (11 percent), and dry tap, nine of 129 (7 percent), being the most common. We conclude that diagnostic thoracocentesis is a clinically valuable procedure if used in conjunction with the patient presentation with an understanding of its limitations for providing a specific etiologic diagnosis. When performed by physicians in training, the number of complications are substantial and the operator often underestimates the degree of patient discomfort. Awareness of the clinical value and complications of thoracocentesis should lead to improved use and safety of this procedure.