OBJECTIVE To describe the relationship of patients' symptoms during therapeutic thoracentesis to pleural pressure (Ppl). METHODS Review of prospectively collected data during 169 therapeutic thoracentesis procedures. METHODS University Hospital in Boston, MA. METHODS One hundred sixty-nine patients who had Ppl measured during therapeutic thoracentesis were included in this study. End-expiratory pressures were measured after the withdrawal of 5 mL of fluid and every 240 mL thereafter until the pressure was lower than -20 cm H(2)O, chest discomfort developed in the patient, or no more fluid could be removed. Patients' symptoms, including chest pain, chest discomfort, and cough were recorded simultaneously. RESULTS There was no correlation between the amount of pleural fluid removed and the development of symptoms. The closing pressures and the total change in Ppl (see the "Materials and Methods" section for definitions), however, were significantly lower in the group of patients who experienced chest discomfort compared to patients who developed cough or were asymptomatic. There was also a trend toward a significantly lower pleural elastance in patients who developed cough compared to that in the other two groups. Additionally, only 22% of patients in whom chest discomfort developed, and 8.6% of patients in whom symptoms did not develop, had potentially dangerous Ppl values (ie, lower than -20 cm H(2)O). CONCLUSIONS The development of chest discomfort is associated with a potentially unsafe drop in Ppl values and should be a sign to terminate thoracentesis. It is not necessary to terminate thoracentesis solely because of the development of cough. Without attention to pleural manometry, a significant percentage of patients may develop potentially dangerous Ppl. Although we recommend pleural manometry with all thoracenteses, when it is not used attention to symptoms remains a valuable surrogate.