BACKGROUND: Emerging studies have shown temperature-mortality association is changing over time, but little is known about the temporal changes of the temperature-morbidity association. OBJECTIVES: We aimed to evaluate the temporal variations in both temperature-respiratory hospitalizations associations and temperature-related attributable risks in Hong Kong. METHODS: We collected 17-year time-series data on daily ambient temperature and emergency hospital admissions for respiratory diseases between 2000 and 2016 in Hong Kong. Quasi-Poisson regression with a time-varying distributed lag nonlinear model was used to estimate the year-specific association between temperature and respiratory hospitalizations [total respiratory, pneumonia, and chronic obstructive pulmonary disease (COPD)] and the year-specific attributable fraction (AF) for heat and cold (defined as above/below the optimum temperature, respectively). RESULTS: Heat-related risks and AFs increased continuously for total respiratory, pneumonia and COPD hospitalizations during the past 17years, respectively. Cold-hospitalization associations and cold-related AFs showed heterogeneous patterns, showing a decreasing trend for pneumonia but a general increasing trend for COPD for both the associations and AFs. The total temperature-related AFs remained stable for total respiratory (p for trend=0.136) and pneumonia (p for trend=0.406), but showed an increasing trend for COPD (p for trend<0.001) from 10% (95% empirical CI: 2%, 17%) in 2000 to 17% (95% empirical CI: 11%, 22%) in 2016. CONCLUSIONS: Our findings indicate an increased susceptibility to heat but a decreased susceptibility to cold for respiratory hospitalizations during the past 17years. The overall temperature-related hospitalization burden for respiratory diseases was generally stable in Hong Kong.