The hemodynamic and electroencephalographic effects of a 60 min infusion of thiopental, given at the rate of 1.25 mg·kg-1·mi-1 (total dose 75 mg/kg), were studied in 10 patients without cardiorespiratory disease undergoing surgery for the removal of large and/or deeply seated arteriovenous malformations. Data on heart rate, arterial, right arterial, pulmonary arterial, and pulmonary capillary wedge pressures, thermodilution cardiac output (expressed as cardiac index), the electroencephalogram, arterial blood gases, and serum thiopental concentrations were collected during a sedated, resting control period, and then every 15 min during drug infusion. Lactated Ringer's solution (total volume 1-2 liters) was infected throughout the study period at rates sufficient to maintain pulmonary capillary wedge pressure at control values. In doses sufficient to render the electroencephalogram isoelectric [t ≃ 30 min, 37.5 mg/kg cumulative dose, serum concentration 51 ± 17 μg/ml (mean ± SD)], drug infusion resulted in significant increases in heart rate (to 116% of control), and decreases in arterial pressure (to 87% of control), stroke volume index (to 87% of control), systemic resistance (84% of control), and both left and right ventricular stroke work indices (66% and 69% of control, respectively). Cardiac index was unchanged (following a transient increase at t = 15 min). There were no changes in pulmonary capillary wedge pressure, pulmonary arterial, pulmonary vascular resistance, or blood gases. A large total dose of thiopental (8-11 gm) was needed to maintain electroencephalogram suppression for the remainder of these 10-20 hr procedures, and emergence was slow (48-72 hr). The resultant prolonged ICU support may have contributed to serious postoperative complications in two patients. These results demonstrate that high concentrations of thiopental are associated with vasodilation and myocardial depression, changes which were similar to those seen in a previous study using methohexital. However, they were of a lesser magnitude than reported with electroencephalographically equivalent doses (2.0 MAC) of isoflurane, suggesting that barbiturates may be hemodynamically preferable if profound electroencephalogram suppression is desired. However, their use is not without risk, and such an anesthetic should be undertaken only with the utmost caution.