Using magnetic resonance spectroscopy, the authors tested whether cerebral concentrations of inhaled anesthetics do not increase proportionately at inspired concentrations exceeding 3% 1) because anesthetics bind to and saturate specific sites in the brain or 2) because anesthetic-induced depression of ventilation limits the increase in alveolar anesthetic partial pressure. New Zealand White rabbits were anesthetized with methohexital, 70% nitrous oxide, and local infiltration of 1% lidocaine. Cerebral concentrations of anesthetic were determined from19F spectra acquired with nuclear magnetic resonance (NMR). Inspired, end-tidal, and arterial anesthetic concentrations, and end-tidal and arterial partial pressure of carbon dioxide were measured. Blood/gas partition coefficients were determined and used to convert arterial anesthetic concentration to partial pressures. In seven spontaneously breathing animals, halothane (1%; n = 5) or isoflurane (0.8%; n = 2) was administered at a constant inspired concentration for 20 min; NMR spectra were acquired between 10 and 20 min. Thereafter, the inspired concentration was increased and the process repeated until apnea occurred. Two additional rabbits were anesthetized with isoflurane and studied similarly but with higher inspired concentrations during mechanical ventilation. In spontaneously breathing animals, ventilatory depression occurred, documented by marked increases in PaCO2, and cerebral concentrations of anesthetic did not increase proportionately at inspired concentrations exceeding 3%. In contrast to an absence of a correlation of inspired and cerebral concentrations during spontaneous ventilation, arterial and cerebral concentrations correlated linearly during both spontaneous and mechanical ventilation (R2> 0.969). These results are consistent with depression of ventilation, rather than binding to specific cerebral sites as an explanation for the nonlinear relationship between cerebral and inspired anesthetic concentrations.