The repressor element 1-silencing transcription (REST) factor is a key regulator of the aging brain's stress response. It is reduced in conditions of stress and Alzheimer's disease (AD), which suggests that increasing REST may be neuroprotective. REST can be measured peripherally in blood plasma. Our study aimed to (1) examine plasma REST levels in relation to clinical and biological markers of neurodegeneration and (2) alter plasma REST levels through a stress-reduction intervention-mindfulness training. In study 1, REST levels were compared across the following four well-characterized groups: healthy elderly (n=65), mild cognitive impairment who remained stable (stable MCI, n=36), MCI who later converted to dementia (converter MCI, n=29) and AD (n=65) from the AddNeuroMed cohort. REST levels declined with increasing severity of risk and impairment (healthy elderly>stable MCI>converter MCI>AD, F=6.35, P<0.001). REST levels were also positively associated with magnetic resonance imaging-based hippocampal and entorhinal atrophy and other putative blood-based biomarkers of AD (Ps<0.05). In study 2, REST was measured in 81 older adults with psychiatric risk factors for AD before and after a mindfulness-based stress reduction intervention or an education-based placebo intervention. Mindfulness-based training caused an increase in REST compared with the placebo intervention (F=8.57, P=0.006), and increased REST was associated with a reduction in psychiatric symptoms associated with stress and AD risk (Ps<0.02). Our data confirm plasma REST associations with clinical severity and neurodegeneration, and originally, that REST is modifiable by a psychological intervention with clinical benefit.