Background <p>People living with dementia may be particularly vulnerable to extreme temperature because of impaired thermoregulation and reduced adaptive capacity. There remains a gap regarding the impacts of extreme temperatures on dementia mortality and the associated mortality burden, especially in northern China.</p> Methods <p>A case time series (CTS) design was employed to evaluate both the cumulative and lag effects, using 21,358 dementia-related deaths recorded in Hebei Province, China, from 2014 to 2022. Daily mean temperature was obtained from the ERA5 reanalysis dataset and spatially matched to residential locations. Conditional Poisson Regression combined with distributed lag non-linear models were applied to estimate the associations. Stratified analyses were performed by age, sex, and marital status. Attributable fractions were calculated to quantify the mortality burden.</p> Results <p>Non-linear associations were observed, with the minimum mortality temperature at 23.4°C. Extreme heat was associated with a significantly increased mortality risk over lag 0-14 (RR=1.71, 95%CI: 1.37-2.14), whereas the effect of extreme cold was not statistically significant (RR=1.42, 95%CI: 0.96-2.09). Heat-related effects were strongest on the day of exposure and persisted for up to 11 days. Elevated heat-related risks were observed across many subgroups, with a significant interaction by marital status. Overall, 20.42% of dementia mortality was attributable to non-optimal temperatures, with extreme heat accounting for a larger proportion than extreme cold (1.39% vs. 0.91%).</p> Conclusions <p>Short-term exposure to extreme heat substantially increased dementia-related mortality in northern China, with both immediate and delayed effects. While extreme cold showed weaker cumulative associations, delayed risks at longer lags suggested that cold-related impacts should not be overlooked.</p>