The authors prospectively investigated the ability of thiopental to decrease neuropsychiatric complications as a consequence of openventricle operations requiring cardiopulmonary bypass. Eighty-nine randomly assigned patients received sufficient thiopental to maintain electroencephalographic silence throughout the period from before atrial cannulation to termination of bypass. These patients received an average of 39.5 mg/kg of thiopental, while 93 control patients received only fentanyl. On the first postoperative day, five thiopental (5.6%) and eight control (8.6%) patients exhibited clinical neuro-psychiatric abnormalities. By the tenth postoperative day, all neuropsychiatric dysfunction had resolved in the thiopental group but persisted in seven (7.5%) control patients (P< 0.025). The incidence of complications was significantly related to calcification of replaced valves, aortic valve replacement, advanced age, and prolonged bypass, but not to low blood pressure during perfusion. The authors believe their data are consistent with embolism as the most important cause of sensory-motor neurologic dysfunction following cardiopulmonary bypass. The data also provide evidence that thiopental in sufficient dosage can reduce the clinical consequences of these events. This is the first demonstration of cerebral protection by a barbiturate in humans.