We examined the changes in cognitive function due to restrictions in daily life during the coronavirus disease 2019 (COVID-19) pandemic in community-dwelling older adults with mild cognitive decline. This was a retrospective, case-control study. The participants include 88 older adults with mild cognitive decline (mean age = 81.0 [standard deviation = 6.5] years) who participated in a class designed to help prevent cognitive decline. This class was suspended from early-March to end of May 2020 to prevent the spread of COVID-19, and resumed in June 2020. We collected demographic and cognitive function test data (Touch Panel-type Dementia Assessment Scale [TDAS]) before and after class suspension and questionnaire data on their lifestyle and thoughts during the suspension. Change in TDAS scores from before and after the suspension was used to divide the participants into decline (2 or more points worsening) and non-decline (all other participants) groups, with 16 (18.2%) and 72 (81.8%) participants in each group, respectively. A logistic regression model showed that the odds ratio (OR) for cognitive decline was lower in participants whose responses were "engaged in hobbies" (OR = 0.07, p = 0.015), "worked on a worksheet about cognitive training provided by the town hall" (OR = 0.19, p = 0.026), and "had conversations over the phone" (OR = 0.28, p = 0.0495). There was a significant improvement in TDAS scores after class was resumed (p < 0.01). A proactive approach to intellectual activities and social ties may be important for the prevention of cognitive decline during periods of restrictions due to COVID-19. We found that cognitive function test scores before class suspension significantly improved after resuming classes. We speculate that continued participation in this class led to positive behavioral changes in daily life during periods of restriction due to COVID-19.