While current arousal level assessments in patients with disorders of consciousness discriminate altered states of consciousness, there are significant limitations in characterizing the transition from one state to another or quantifying the frequent arousal level fluctuations observed in a patient. Here, we identified a repeated, temporally discrete, dynamical pattern evident in the recovery of consciousness from anesthesia and brain injury coma models in rodents. We prospectively validated these features we label "Arousal Units" (AU) in neonatal humans recovering from static hypoxic injuries and senior patients emerging from anesthesia indicating their generalizability. The AUs lawfully link changes in spectral power and breathing frequency and reliably associate with motor changes. Distinctive cortical patterns within AUs can be transformed into arousal indices, determining arousal levels. The reliability of these events is demonstrated across intact and brain-injured states and translates to the human brain; extracting these stereotyped dynamics could aid anesthesia monitoring, tracking coma recovery, and identifying cognitive motor dissociation.