Behavioral reports have traditionally been the gold standard for evaluating the presence of consciousness. However, it is becoming clear that consciousness can be present even in the absence of overt behavior and in unresponsive subjects. I will present neurophysiological evidence supporting the presence of consciousness in dissociated states from several domains. Measures of cortical integration and differentiation have recently proven to be the most reliable marker of consciousness irrespective of behavior and have been validated in a large number of different conditions. The most common dissociation between consciousness and behavior occurs every night during dreaming sleep. Recent work using both within-state, no-task paradigms and TMS-EEG shows that consciousness can be present during non REM sleep when the front of the brain shows high amplitude slow waves, as long as a posterior cortical hot zone is activated. Studies using different anesthetics have also shown that fully unresponsive subjects anesthetized with ketamine (as compared to propofol or xenon) retrospectively report intense dreams, which are again associated with high complexity responses to TMS, despite the occurrence of slow waves. High complexity responses can also be observed in about 20% of patients in a vegetative state suggesting, in line with previous findings using active paradigms, that a number of completely unresponsive patients may retain consciousness. Finally, a number of studies in healthy awake volunteers have emphasized frequent dissociations between consciousness and task-related cognitive functions. Overall, recent findings show that the anatomical neural correlates of consciousness are primarily localized to a posterior cortical hot zone that includes sensory areas, rather than to a fronto-parietal network involved in task monitoring and reporting. I will end by discussing promising avenues of future research.