General anesthetic agents are thought to induce loss-of-consciousness (LOC) and enable pain-free surgery by acting on the endogenous brain circuitry responsible for sleep-wake cycling. In clinical use, the entire CNS is exposed to anesthetic molecules with LOC and amnesia usually attributed to synaptic suppression in the cerebral cortex and immobility and analgesia to agent action in the spinal cord and brainstem. This model of patch-wise suppression has been challenged, however, by the observation that all functional components of anesthesia can be induced by focal delivery of minute quantities of GABAergic agonists to the brainstem mesopontine tegmental anesthesia area (MPTA). We compared spectral features of the cortical electroencephalogram (EEG) in rats during systemic anesthesia and anesthesia induced by MPTA microinjection. Systemic administration of (GABAergic) pentobarbital yielded the sustained, δ-band dominant EEG signature familiar in clinical anesthesia. In contrast, anesthesia induced by MPTA microinjection (pentobarbital or muscimol) featured epochs of δ-band EEG alternating with the wake-like EEG, the pattern typical of natural non-rapid-eye-movement (NREM) and REM sleep. The rats were not sleeping, however, as they remained immobile, atonic and unresponsive to noxious pinch. Recalling the paradoxical wake-like quality the EEG during REM sleep, we refer to this state as “paradoxical anesthesia”. GABAergic anesthetics appear to co-opt both cortical and spinal components of the sleep network via dedicated axonal pathways driven by MPTA neurons. Direct drug exposure of cortical and spinal neurons is not necessary, and is probably responsible for off-target side-effects of systemic administration including monotonous δ-band EEG, hypothermia and respiratory depression. Significance statementThe concept that GABAergic general anesthetic agents induce loss-of-consciousness by substituting for an endogenous neurotransmitter, thereby co-opting neural circuitry responsible for sleep-wake transitions, has gained considerable traction. However, the electroencephalographic (EEG) signatures of sleep and anesthesia differ fundamentally. We show that when the anesthetic state is generated by focal delivery of GABAergics into the mesopontine tegmental anesthesia area (MPTA) the resulting EEG repeatedly transitions between delta-wave-dominant and wake-like patterns much as in REM-NREM sleep. This suggests that systemic (clinical) anesthetic delivery, which indiscriminately floods the entire cerebrum with powerful inhibitory agents, obscures the sleep-like EEG signature associated with the less adulterated form of anesthesia obtained when the drugs are applied selectively to loci where the effective neurotransmitter substitution actually occurs.