Over the past decade, Brazil has experienced and continues to be impacted by extreme climate events. This study aims to evaluate the association between daily average temperature and mortality from respiratory disease among Brazilian elderlies. A daily time-series study between 2000 and 2017 in 27 Brazilian cities was conducted. Data outcomes were daily counts of deaths due to respiratory diseases in the elderly aged 60 or more. The exposure variable was the daily mean temperature from Copernicus ERA5-Land reanalysis. The association was estimated from a two-stage time series analysis method. We also calculated deaths attributable to heat and cold. The pooled exposure–response curve presented a J-shaped format. The exposure to extreme heat increased the risk of mortality by 27% (95% CI: 15–39%), while the exposure to extreme cold increased the risk of mortality by 16% (95% CI: 8–24%). The heterogeneity between cities was explained by city-specific mean temperature and temperature range. The fractions of deaths attributable to cold and heat were 4.7% (95% CI: 2.94–6.17%) and 2.8% (95% CI: 1.45–3.95%), respectively. Our results show a significant impact of non-optimal temperature on the respiratory health of elderlies living in Brazil. It may support proactive action implementation in cities that have critical temperature variations.