White matter lesions (WML) on magnetic resonance imaging (MRI) brain scans are associated with ageing. They are unrelated to specific disorders, and their impact on cognitive and other brain functions is poorly characterized. Pathological studies often omit systematic survey of WML because of the need to study multiple full coronal tissue blocks, and uncertainty over the significance of lesions identified in periventricular and deep subcortical regions. Post-mortem MRI provides a means of mapping WML but the sensitivity and specificity of the method are unresolved. In this study post-mortem MRI of WML in fixed brain slices was compared with pathology in 33 brains donated to the Medical Research Council Cognitive Function and Ageing Study (MRC CFAS). This study shows that MRI detection of WML was less sensitive than pathology: periventricaular lesions (PVL) sensitivity = 95% (87-99%), specificity = 71% (44-90%); deep subcortical lesions (DSCL) sensitivity = 86% (79-93%), specificity = 80% (72-88%). False negative MRI was associated with milder pathology, but lesions detected by myelin attenuation alone showed both microglial and endothelial activation. Therefore post-mortem MRI of formalin-fixed brain slices is a reliable method to obtain systematic data on the severity and distribution of cerebral white matter disease, and appears to detect those WML most likely to have clinical impact.