Objectives : Alzheimer's disease (AD) and mild cognitive impairment (MCI) are often accompanied by neuropsychiatric symptoms (NPS; e.g. depression/apathy/irritability) causing challenges for people living with dementia/caregivers and predicting worse disease progression. Accurately assessing NPS is critical to research on AD/MCI. However, there are limitations to both self-reports and clinician evaluations; the field often relies on informants to assess NPS. Informants' perception of NPS are influenced by disease and caregiver factors that may lead to biased assessments. We aimed to assess the relationship between participants self-reported affective states (valence/arousal) and informant-reported NPS. Methods : Data from a double-blinded intervention design (primarily testing neurostimulation's effect on NPS) were used to examine the relationship between participant-reported affective states and informant-reported NPS over 1 month. Forty participants (24 females) with MCI and NPS (mean age = 71.7, SD = 7) were enrolled along with informants (primarily spouses/partners) who regularly interact with participants. NPS assessment occurred weekly and at pre- and post-intervention, and participant-reported affective states were assessed at 14 timepoints. Results : Generalized Estimating Equations showed that participant levels of arousal, but not valence, were significantly related to corresponding informant-reported NPS at weekly (arousal: B = -0.59, SE = 0.27, Wald's χ 2 = 4.61, p =.032; valence: B = 0.17, SE = 0.19, Wald's χ 2 = 0.80, p =.37) and pre-/post- (arousal: B = -4.00, SE = 1.58, Wald's χ 2 = 6.42, p =.011; valence: B = -3.34, SE = 1.80, Wald's χ 2 = 3.43, p =.06) assessments. Conclusion : The findings indicate that informant-reported NPS may be more strongly influenced by arousal, and informants may be less attuned to valence in people living with MCI.