The plasma neurofilament light chain (NfL), an axonal cytoskeleton protein, increases in Alzheimer disease and was therefore proposed as a blood-based biomarker of the disease. Tensor-based morphometry (TBM) is an MR based modality that identifies local volume changes in the brain. Herein, we aimed to investigate whether plasma NfL measures can predict TBM findings derived from temporal lobe of brain in a one-year follow-up and which biomarker can predict cognitive function. A total of 480 participants with Alzheimer's disease (AD), mild cognitive impairment (MCI), and normal cognition (CN) were found eligible for inclusion from The Alzheimer's Disease Neuroimaging Initiative (ADNI) database. There was a significant negative association between plasma NfL and TBM only when all subjects were pooled together at baseline (β = -0.139, P= 0.004). After one-year follow-up, 30 subjects with MCI converted to AD (MCI-AD) and others remained unchanged (CN, MCI, AD). Plasma NfL levels elevated significantly only in MCI group after one year (P<0.001). We found a significant reduction in TBM measurements at first-year compared to baseline in all groups (P<0.001 for all groups). Additionally, TBM average change rate was significantly higher in MCI-AD and AD groups (P<0.001 for both); however, plasma NfL average change rate was not significantly different between groups. TBM was significantly correlated with MMSE, MoCA, ADAS-11 and ADAS-13 scores in both MCI and AD patients at baseline and after one year, whereas plasma NfL was not. Overall, our findings indicate that plasma NfL is not reliably associated with TBM, and is less effective and sensitive than TBM in predicting dementia progression and cognitive performance. Hence, TBM reduction is not reflected in plasma NfL increment after one year follow-up.