Theoretically, precipitation extremes will increase at a rate of 6-7% with temperature increasing, namely the Clausius-Clapeyron relationship. However, many gauge observations suggest a peak structure of the relationship between precipitation extremes and atmospheric temperature, deviating from the Clausius-Clapeyron relationship. In this study, a comprehensive investigation about the temperature dependence of precipitation extremes (hourly, daily, and event-based) across China is implemented. The results confirm the widespread existence of the peak structure for daily and hourly precipitation extremes and show that (1) there is a generally positive spatial correlation between the precipitation extremes at the peak and temperature at the peak, and this scaling rate is close to the C-C rate; (2) the scaling of event-based extremes for precipitation amount with temperature follows a similar pattern to the daily precipitation extremes while the event-based precipitation intensity does not show a peak structure; (3) the decrease of rain duration is the main cause for the peak structure of the rain amount scaling.