Objective To understand the impact of daily average temperature on years of life lost (YLL) due to non-accidental and cardiovascular and cerebrovascular disease deaths in seven mortality surveillance sites in Xinjiang Uyghur Autonomous Region, and to provide scientific evidence for reducing the loss of life expectancy caused by daily average temperature. Methods Mortality surveillance data for residents reported from January 1, 2015, to December 31, 2019, were collected from seven mortality surveillance sites (Tianshan district, Urumqi city; Karamay district, Karamay city; Yizhou district, Hami city; Korla city; Hotan county; Artux city; and Tacheng city) in Xinjiang Uyghur Autonomous Region through the China Center for Disease Control and Prevention’s Population Mortality Information Reporting System. Air quality data from the National Environmental Expertise Service Platform and meteorological data from the Climate Center of the Xinjiang Uyghur Autonomous Region Meteorological Bureau for the same period were also collected. A distributed lag non-linear model (DLNM) was used to quantitatively analyze the impact of daily average temperature on residents' YLL and the lag effect. Subgroup analyses were conducted for different genders and age groups to identify sensitive populations. Results A total of 57 094 non-accidental deaths and 26 8484 cardiovascular and cerebrovascular disease deaths were reported in the seven mortality surveillance sites in Xinjiang Uyghur Autonomous Region from 2015 to 2019. The daily average YLL values for non-accidental deaths and cardiovascular and cerebrovascular disease deaths were (795.18±210.15) years and (332.15±111.58) years, respectively. The YLL values for both non-accidental and cardiovascular and cerebrovascular disease deaths showed a certain degree of seasonal fluctuation with daily average temperature from 2015 to 2019 in the seven surveillance sites. The trends in YLL for both causes of death were opposite to the temperature trend, with the highest YLL values occurring in January and December. DLNM analysis showed that the effect of high temperature (>22.57 °C) on the overall YLL was greatest on the day of exposure. For every 1°C increase in daily average temperature, the YLL due to non-accidental and cardiovascular and cerebrovascular disease deaths increased by [16.021 (95%CI: 10.125–21.917)] years and [5.821 (95%CI: 2.579–9.063)] years, respectively. The effect of low temperature (<−1.04°C) on overall YLL had a lag effect, with the maximum effect occurring at lag 2 days for non-accidental deaths and lag 5 days for cardiovascular and cerebrovascular disease deaths. For every 1°C decrease in daily average temperature, the YLL due to non-accidental deaths at lag 2 days and cardiovascular and cerebrovascular disease deaths at lag 5 days increased by [9.126 (95%CI: 0.872–17.053)] years and [2.739 (95%CI:0.983～5.086)] years, respectively. Subgroup analysis showed that the effect of high temperature (>22.57 °C) on YLL was greatest on the day of exposure for both males and females. For every 1°C increase in daily average temperature, the YLL due to non-accidental deaths in males on the day of exposure increased by [7.559 (95%CI: 3.161–11.956)] years, and the YLL due to non-accidental and cardiovascular and cerebrovascular disease deaths in females on the day of exposure increased by [8.462 (95%CI: 4.994–11.930)] years and [3.589 (95%CI: 1.862–5.315)] person-years, respectively. The effect of low temperature (<−1.04 °C) on YLL had a lag effect for both males and females, with the maximum effect occurring at lag 3 days for non-accidental deaths in males and at lag 2 days for non-accidental and cardiovascular and cerebrovascular disease deaths in females. For every 1 °C decrease in daily average temperature, the YLL due to non-accidental deaths at lag 3 days in males increased by [6.831 (95%CI: 0.657–13.625)] years, and the YLL due to non-accidental and cardiovascular and cerebrovascular disease deaths at lag 7 days and lag 6 days in females increased by [4.530 (95%CI: 0.235–7.864)] years and [1.824 (95%CI: 0.189–3.382)] years, respectively. The effect of high temperature (>22.57 °C) on YLL was greatest on the day of exposure for both residents aged <65 years and ≥65 years. For every 1°C increase in daily average temperature, the YLL due to non-accidental deaths on the day of exposure increased by [8.487 (95%CI: 3.398–13.577)] years for residents aged <65 years, and the YLL due to non-accidental and cardiovascular and cerebrovascular disease deaths on the day of exposure increased by [7.524 (95%CI: 4.949–10.099)] person-years and [3.690 (95%CI: 1.884–5.497)] person-years, respectively, for residents aged ≥65 years. The effect of low temperature (< −1.04 °C) on YLL had a lag effect for both residents aged <65 years and ≥65 years, with the maximum effect occurring at lag 3 days for non-accidental deaths in residents aged <65 years and on the day of exposure for non-accidental deaths and at lag 4 days for cardiovascular and cerebrovascular disease deaths in residents aged ≥65 years. For every 1 °C decrease in daily average temperature, the YLL due to non-accidental deaths at lag 3 days in residents aged <65 years increased by [4.235 (95%CI: 1.583–9.648)] years, and the YLL due to non-accidental deaths on the day of exposure and cardiovascular and cerebrovascular disease deaths at lag 4 days in residents aged ≥65 years increased by [8.277 (95%CI: 1.885–14.668)] years and [2.855 (95%CI: 0.273–5.192)] years, respectively. Conclusions Both high and low temperatures increase YLL in residents of the seven mortality surveillance sites in Xinjiang Uyghur Autonomous Region, with a lag effect. Females and residents aged ≥65 years are sensitive populations.