Aging is associated with declining cognitive function in cross-sectional studies, but few studies have tested for longitudinal change in hippocampal structure or integrity. Cellular changes following aging, such as neuronal degeneration and synaptic loss, can be investigated using quantitative MRI measurements such as longitudinal relaxation time (T1), magnetization transfer ratio (MTR), fractional anisotropy (FA) and mean diffusivity (MD). T1 is related to brain tissue water content, while MTR measures the quantity of bound water protons present in the brain tissue. FA and MD are scalar indices obtained from the diffusion tensor (DT), with FA signifying the directionality dependence of water molecules when subjected to cellular boundaries within a tissue, and MD representing overall magnitude of water diffusion. MTR and FA are reduced and T1 and MD are increased in many pathologies, attributed to demyelination. We tested whether there were coupled changes in hippocampal volume, T1, MTR, FA and MD and three broad cognitive domains (Verbal Memory, Working Memory, and Speed) in a large sample of community-dwelling non-demented adults at 73 and 76 years of age.A large cohort of human participants from Lothian Birth Cohort 1936 completed tests of Verbal Memory (WMS-III^UK Logical Memory and Verbal Paired Associates), Working Memory (WMS-III^UK Spatial Span, WAIS-III^UK Digit Span Backward and Letter-Number Sequencing) and Speed (WAIS-III^UK Digit-Symbol Substitution, Symbol Search, and tests of Choice Reaction Time and Inspection Time) at mean age 73 (n = 731) and 76 (n = 488) years¹. From this cohort, 655 underwent brain MRI scans at age 73 years and 469 were scanned at age 76 years on a 1.5T GE clinical scanner as described previously². Hippocampi were segmented on T1-weighted volumes using an age-relevant template and FSL tools plus manual editing. T1 maps were obtained from FSPGR sequences and MTR maps from signal intensities with and without an MT pulse^2,3; these maps were previously registered to T2-weighted volumes using FSL-FLIRT. DT-MRI data were preprocessed using FSL (http://www.ndcn.ox.ac.uk/divisions/fmrib/), constructing FA and MD maps. Hippocampal binary masks were mapped (Fig. 1) so that median values of T1, MTR, FA and MD within the hippocampal masks were obtained. As baseline hippocampal measures were significantly correlated with multiple cognitive measures⁴, we used latent change score modeling to model cross-wave change in each of the hippocampal measures alongside change in latent factors of Verbal Memory, Working Memory, and Speed, each indicated by multiple cognitive scores.Hippocampal volume declined from age 73 to 76 (r = −0.389, p < 0.001), as did FA (weakly: r = −0.057, p < 0.001), and MTR (though non-significantly; r = −0.173, p = 0.113); MD increased with age (r = 0.657, p < 0.001), although unexpectedly T1 showed significant age-related decline (r = −0.216, p < 0.001). We tested all associations between levels and changes of the hippocampal measures and levels and changes of the three cognitive measures. There were few correlations between baseline hippocampal measures and cognitive change, with the exception of those shown in Fig. 2: higher baseline MD was correlated with steeper cognitive decline in all three factors. Unexpectedly, hippocampal MD showed (marginally significant) coupled change with Working Memory, such that individuals with higher MD increases (i.e. more apparent deterioration in tissue integrity) also tended to show increases in Working Memory (r = 0.178, p = 0.041).We found that hippocampal integrity assessed using MD was significantly associated with all measures of cognitive ability investigated here. The age related decline of T1 may result from partially restored mechanisms of plasticity and altered hippocampal neuronal ensembles⁵, although inaccurate partial volume error on small hippocampal structures is possible.