General anesthetics induce a temporary and reversible state of suppressed wakefulness, impaired responsiveness to environmental stimuli, immobility, and amnesia. Yet how the brain recovers responsiveness, consciousness, and cognition after perturbation by anesthesia is not well characterized. This thesis aims to improve understanding of the neurobiological mechanisms that underlie transitions between the anesthetized state and wake. Under fixed, intermediate doses of anesthesia, individuals fluctuate spontaneously between states of unresponsiveness and wake. How densely or poorly consolidated these state transitions are depends on which anesthetic is administered. To test the hypothesis that the orexin system, integral to sleep-wake state consolidation, regulates arousal state fluctuations under anesthesia, we use a combination of behavioral, electrophysiological, genetic, pharmacological, and circuit mapping techniques. We find that the actions of orexins at orexin receptor-2 (OXR2) are necessary and sufficient for consolidation of behavioral and cortical activity states under anesthesia. Further, using two-photon imaging in vivo, we demonstrate that the activation of OXR2 locally in the cortex evokes synchronous bursting activity in layer 5 neurons. This finding provides preliminary support for a mechanism by which orexins act directly on the cortex to exert global arousal state-stabilizing actions. These studies identify OXR2-selective drugs as potential therapeutic adjuncts in the perioperative period. In a complimentary set of experiments, we propose a novel model for anesthetic emergence. In ten separate sessions, we expose mice to anesthesia and measure time to return of responsiveness. Despite identical anesthetic exposures and a genetically homogeneous population, we find emergence times span two orders of magnitude for both individuals and the population. We test how well the predictions of two mathematical models quantitatively explain this wide variability. We find that a classical pharmacokinetic-pharmacodynamic model can recapitulate population-level features of emergence but fails to capture the variability in emergence times we observe in individuals. Instead, a neuronal dynamics model, which incorporates the inertial tendency of the brain to resist arousal state transitions, can accurately capture all population- and individual-level features of emergence. Altogether, the results of this thesis advance our current understanding of factors beyond pharmacokinetics and pharmacodynamics that influence exit from the state of anesthesia.