Adaptations in brain communication are associated with multiple pain disorders and are hypothesized to promote the transition from acute to chronic pain. Despite known increases in brain synaptic activity, it is unknown if and how changes in pathways and networks contribute to persistent pain. A tunable rat model that induces transient or persistent temporomandibular joint pain was used to characterize brain network and subcircuit changes when sensitivity is detected in both transient and persistent pain groups and later when sensitivity is present only for the persistent pain group. Brain activity was measured by F-FDG positron emission tomography imaging and used to construct intersubject correlation networks; network connectivity distributions, diagnostics, and community structure were assessed. Activation of subcircuits was tested by structural equation modeling. Findings reveal differences in the brain networks at day 7 between the persistent and transient pain groups, a time when peripheral sensitivity is detected in both groups, but spontaneous pain occurs only in the persistent pain group. At day 7, increased (P ≤ 0.01) clustering, node strength, network segregation, and activation of prefrontal-limbic pathways are observed only in the group that develops persistent pain. Later, increased clustering and node strength are more pronounced with persistent pain, particularly within the limbic system, and decrease when pain resolves. Pretreatment with intra-articular etanercept to attenuate pain confirms that these adaptations are associated with pain onset. Results suggest that early and sustained brain changes can differentiate persistent and transient pain, implying they could be useful as prognostic biomarkers for persistent pain and in identifying therapeutic targets.