STUDY OBJECTIVES: To assess the impact of routine thoracentesis on diagnostic assessment and therapeutic measures in patients with clinically documented pleural effusions. DESIGN AND SETTING: Prospective, 1-year, three-center study in medical ICU (MICU) patients with physical and radiographic evidence of pleural effusion. PATIENTS: Of 1,351 patients admitted to three MICUs during the study period, 113 patients had physical and radiographic evidence of pleural effusion, yielding an annual incidence of 8.4%. INTERVENTION: Routine thoracentesis in 82 patients without contraindications to thoracentesis. MEASUREMENTS AND RESULTS: Twenty patients (24.4%) had a transudate, 35 patients (42.7%) had an infectious exudate (parapneumonic, n = 21; empyema, n = 14), and 27 patients (32.9%) had a noninfectious exudate. Laboratory parameters including the leukocyte count, the neutrophil percentage in pleural fluid, and the fluid/serum protein and lactate dehydrogenase ratios differed significantly among the three groups. Thoracentesis yielded improvements in the diagnosis and/or treatment in 46 patients (56%): the presumptive (prethoracentesis) diagnosis was changed in 37 patients (32 patients with certain benefit and 5 patients with probable benefit from thoracentesis), of whom 27 patients received a change in treatment based on the new diagnosis; 9 other patients received a change in treatment although the diagnosis remained the same. The only complications were pneumothorax in six patients (7%), all with a favorable outcome after drainage. CONCLUSION: Infection was the main cause of pleural effusions detected based on physical and radiographic findings in our MICU population. Routine thoracentesis proved a simple and safe means of improving the diagnosis and treatment.