Medication overuse headache (MOH) is a prevalent secondary headache, bringing heavy economic burden and neuropsychological damage. Neuroimaging studies on the disease reported divergent results. To merge the reported neuroimaging alterations in MOH patients and explore a pathophysiological mechanism of this disorder. A meta-analytic activation likelihood estimation (ALE) analysis method was used. We systematically searched English and Chinese databases for both morphological and functional neuroimaging studies published before Nov 18, 2021. Reported altered brain regions and the stereotactic coordinates of their peaks were extracted and pooled by GingerALE using Gaussian probability distribution into brain maps, illustrating converged regions of alteration among studies. We identified 927 articles, of which five studies on gray matter changes, using voxel-based morphometry (VBM) were eventually included for ALE analysis, with 344 subjects and 54 coordinates put into GingerALE. No functional magnetic resonance imaging (fMRI) or positron emission topography (PET) studies were included for pooling. Compared with healthy controls (HCs), MOH featured increased gray matter density in midbrain, striatum, cingulate, inferior parietal cortex and cerebellum (P < 0.001 uncorrected), whereas decreased gray matter density in orbitofrontal cortex (P < 0.05, family-wise error), frontal, insular and parietal cortices (P < 0.001 uncorrected). Withdrawal of analgesics led to decreased gray matter density in superior temporal gyrus, cuneus, midbrain and cerebellum (P < 0.001 uncorrected). This meta-analysis confirmed that medication overuse headache is associated with morphologic alteration in the reward system, the prefrontal cortex and a reversible modification in the pain network. Further functional imaging paradigms and longitudinal studies are required for a more definite conclusion and a causal mechanism.