Brain functional connectivity undergoes dynamic changes from the awake to unconscious states. However, how the dynamics of functional connectivity patterns are linked to consciousness at the behavioral level remains elusive. Here we acquired resting-state functional magnetic resonance imaging (rsfMRI) data during wakefulness and graded levels of consciousness in rats. Data were analyzed using a dynamic approach combining the sliding-window method and k-means clustering. Our results demonstrate that whole-brain networks contain several quasi-stable patterns that dynamically recurred from the awake state into anesthetized states. Remarkably, two brain connectivity states with distinct spatial similarity to the structure of anatomical connectivity were strongly biased toward high and low consciousness levels, respectively. These results provide compelling neuroimaging evidence linking the dynamics of whole-brain functional connectivity patterns and states of consciousness at the behavioral level.