Human prion diseases are fatal neurodegenerative disorders caused by misfolding of the prion protein. There are no useful biomarkers of disease progression. Cerebral cortex spongiform change, one of the classical pathological features of prion disease, resolves in prion-infected transgenic mice following prion protein gene knockout. We investigated the cross-sectional, longitudinal and post-mortem cerebral magnetization transfer ratios as a surrogate for prion disease pathology. Twenty-three prion disease patients with various prion protein gene mutations and 16 controls underwent magnetization transfer ratio and conventional magnetic resonance imaging at 1.5 T. For each subject, whole-brain, white and grey matter magnetization transfer ratio histogram mean, peak height, peak location, and magnetization transfer ratio at 25th, 50th and 75th percentile were computed and correlated with several cognitive, functional and neuropsychological scales. Highly significant associations were found between whole brain magnetization transfer ratio and prion disease (P < 0.01). Additionally, highly significant correlations were found between magnetization transfer ratio histogram parameters and clinical, functional and neuropsychological scores (P < 0.01). Longitudinally, decline in the Clinician's Dementia Rating scale was correlated with decline in magnetization transfer ratio. To investigate the histological correlates of magnetization transfer ratio, formalin-fixed cerebral and cerebellar hemispheres from 19 patients and six controls underwent magnetization transfer ratio imaging at 1.5 T, with mean magnetization transfer ratio calculated from six regions of interest, and findings were followed-up in six variant Creutzfeldt-Jakob disease cases with 9.4 T high-resolution magnetization transfer imaging on frontal cortex blocks, with semi-quantitative histopathological scoring of spongiosis, astrocytosis and prion protein deposition. Post-mortem magnetization transfer ratios was significantly lower in patients than controls in multiple cortical and subcortical regions, but not frontal white matter. Measurements (9.4 T) revealed a significant and specific negative correlation between cortical magnetization transfer ratios and spongiosis (P = 0.02), but not prion protein deposition or gliosis. The magnetic resonance imaging measurement of magnetization transfer ratios may be an in vivo surrogate for spongiform change and has potential utility as a therapeutic biomarker in human prion disease.