Aim: To quantitatively investigate the brain structure of patients with fibromyalgia (FMS) using MRI and the advanced analytical technique of Statistical Parametric Mapping (SPM). Methods: 30 women with primary FMS, 9 taking no medications, and 29 age- and gender-matched healthy controls underwent 1.5 Tesla T1- and T2-weighted MRI brain scans comprising 4-mm trans-axial slices. All had a blinded manual tender point examination and completed standard clinicalvariable questionnaires (FIQ, MPQ, STAI, CES-D). Using the SPM99 and SPM2 software packages, all MRI images were transformed to a standard anatomical space, scaled and subjected to voxel-by-voxel statistics. Voxels with uncorrected P < 0.001 were thereby identified. Results: Regression analysis of T2 scans revealed increased signal (corrected cluster P = 0.031) in the posterior orbito-frontal regions in FMS versus control subjects, adjusted for age. Regression of T2 scans within the FMS group alone detected with increasing pain duration (0.5–45 years) a widespread patchy increase in white matter signal, significant in the dorsal right insula (corrected cluster P = 0.001) and right dorsal prefrontal lobe (corrected cluster P = 0.028) adjusted for age, and with increasing physical impairment last week a reduction in medial posterior cerebellar signal (corrected cluster P < 0.0005) adjusted for age. No relationship with any other clinical variable was detected, including psychological and drug usage. Conclusions: Our results suggest that sub-visual cerebral structural changes occur in FMS unrelated to psychological distress and medications, and may provide a patho-anatomical basis for its associated cold allodynia and impaired working memory