High dose barbiturate therapy has been used clinically for about 10 years. However, the hemodynamic consequences of such therapy have not been well defined in humans. We therefore examined the effects of a brief (30-minute), high dose (0.6 mg/kg/minute or 18 mg/kg total) infusion of pentobarbital in nine otherwise healthy patients (aged 20 to 30) scheduled to undergo operation for the removal of large or deeply seated arteriovenous malformations. Monitored variables included intravascular pressures, cardiac output, arterial and mixed venous blood gases and hematocrit, and the electroencephalogram (EEG). After a brief rest period and the collection of control data, pentobarbital infusion was begun, muscle relaxants were given, and normocarbia was maintained by mask ventilation. Because our primary intent was to examine the effects of the drug on the heart and arterial circulation, lactated Ringer's solution was infused continuously to keep pulmonary capillary wedge pressure (PCWP) at control values in an attempt to keep constant the ventricular 'preload' (although PCWP is only an approximation of the true preload). Drug infusion resulted in progressive EEG suppression ending in a pattern of deep burst suppression (1 to 3 bursts/minute) at t=30 minutes, with measured plasma phenobarbital concentrations (in four patients) of 34 ± 4 μg/ml (mean ±SD). There were no changes in PaO2, PaCO2, or pH, nor were there any changes in PCWP. The cardiac index did not change. This stability was the result of an increase in heart rate (from 72 ± 12 base line to 86 ± 10 beats/minute at t=30 minutes), which acted to compensate for a decrease in calculated stroke volume index (SVI) from 61.8 ± 17.2 control to 47.4 ± 9.1 ml/beat/m2 at t=30 minutes). Mean arterial pressure decreased significantly (from 84 ± 8 to 66 ± 9 mm Hg at t=30 minutes), a change that was due entirely to a reduction in systemic vascular resistance (SVRI). Pulmonary vascular resistance did not change. There was a small but significant dilutional reduction in hematocrit (Hct) from 40 ± 3 to 37 ± 3% at t=30 minutes. This occurred with no change in PCWP (in spite of the infusion of >1 litre of fluid), suggesting that the pentobarbital infusion produced some degree of 'venodilation'. These observations all indicate that pentobarbital - like the 'ultrashort-acting' drugs thiopental and methohexital - is both a myocardial depressant (a decrease in SVI with unchanging PCWP) and a vasodilator (a decrease in SVRI and evidence for venodilation). They also suggest that hemodynamic safety requires that PCWP (and/or right atrial pressures) be maintained at or above 'control' values during drug administration. Furthermore, factors that might prevent the 'compensatory' increase in heart rate (e.g., heart disease, β-blockers) may magnify any drug-induced hypotension by allowing the cardiac index to decrease.