Although assessing a subject’s capacity for consciousness is commonly based on an input-output paradigm, clinical and physiological evidence advocate the development of brain-based metrics that are independent on both sensory inputs and motor outputs. As an attempt in this direction, we devised an empirical measure of complexity derived from the electroencephalographic (EEG) response to a direct cortical perturbation with transcranial magnetic stimulation (TMS), the perturbational complexity index (PCI). Based on theoretical postulates, PCI gauges the conjoint presence of integration and information in the human brain, independently of sensory inputs and motor outputs. In a preliminary series of experiments, PCI was tested on TMS-evoked potentials recorded in healthy subjects during wakefulness, dreaming, NREM sleep, and different levels of sedation induced by different anesthetic agents, as well as in patients who emerged from coma and attained a stable diagnosis. These experiments show that PCI allows a reliable assessment of the level of consciousness at the single-subject level and prompt further validation toward the development of a diagnostic test. In parallel, elucidating the mechanisms by which brain complexity collapses and recovers following brain injury may provide novel insight on the physiopathology and the treatment of disorders of consciousness.