Although interest in assessing the impacts of hot temperature and mortality in Hong Kong has increased, less evidence on the effect of cold temperature on mortality is available. We examined both the effects of heat and cold temperatures on daily mortality in Hong Kong for the last decade (2002-2011). A quasi-Poisson model combined with a distributed lag non-linear model was used to assess the non-linear and delayed effects of temperatures on cause-specific and age-specific mortality. Non-linear effects of temperature on mortality were identified. The relative risk of non-accidental mortality associated with cold temperature (11.1 °C, 1st percentile of temperature) relative to 19.4 °C (25th percentile of temperature) was 1.17 (95% confidence interval (CI): 1.04, 1.29) for lags 0-13. The relative risk of non-accidental mortality associated with high temperature (31.5 °C, 99th percentile of temperature) relative to 27.8 °C (75th percentile of temperature) was 1.09 (95% CI: 1.03, 1.17) for lags 0-3. In Hong Kong, extreme cold and hot temperatures increased the risk of mortality. The effect of cold lasted longer and greater than that of heat. People older than 75 years were the most vulnerable group to cold temperature, while people aged 65-74 were the most vulnerable group to hot temperature. Our findings may have implications for developing intervention strategies for extreme cold and hot temperatures.