Although the depressive effect of sevoflurane on ventilation has been reported, its potency and mode of actio on the neural respiratory activity is still unclear. Therefore, the effects of sevoflurane on the phrenic nerve discharge and the respiratory timing were compared with those of halothane. The efferent activity of the phrenic nerve was recorded from decerebrate, un-anesthetized and artificially vend: lated cats, and its power spectrum was calculated. The inspiratory and expiratory periods were measured. Sevoflurane and halothane of the doses of 0.5-1.5 MAC were inhaled for 15 min. With 0.5 MAC, sevoflurane decreased the total power and two dominant spectral components of the high-frequency oscillation and medium-frequency oscillation in the power spectrum. With the same MAC dose, halothane had a greater depressive effect in a normocapnic condition with the vagus nerves being intact. In a state of hypercapnia or after vagotomy, the effect of halothane was considerably attenuated whereas that of sevoflurane remained unaltered. Halothane increased the neural respiratory rate much more than sevoflurane in both normocapnic and hypercapnic states. Vagotomy significantly weakened the effect of halothane to increase the respiratory rare bur did not modify the effect of sevoflurane. With 1.0-1.5 MAC, both anesthetics severely decreased the phrenic power spectra and the potency difference became indistinct. The present findings demonstrate that sevoflurane has a weaker depressive effect on the respiratory nerve discharge and a smaller effect on the neural respiratory rare than halothane when the effects of 0.5 MAC were compared. This may be due to the lesser effect of sevoflurane on the vagal mediated and CO2-related mechan isms which modulate the global outputs of the central respiratory control system.