The central and eastern part of China experienced a record-breaking extreme high temperature in 2013, which caused serious losses to the lives of residents and the social economy. We used daily maximum temperature from CN05.1 and 17 global climate model data participating in CMIP5 to research the risk changes of such extreme temperature as well as extreme temperature with long return period (20-year, 50-year and 100-year)under 1.5℃ and 2℃ global warming levels. A bias-corrected method was applied to the simulated maximum daily temperature of CMIP5 models. Results show that the bias correction can significantly reduce the biases of the simulated maximum temperature, the risk of occurrence of extreme high temperature days in 2013 will increase by 3.0/6.1 times under 1.5℃/2℃, the intensity of extreme temperature will increase by 5.6/12.6 times at 1.5℃/ 2℃. For extreme high temperatures with different return periods, the rarer high temperatures have larger increase of risk in the future, and the risk of days of extreme temperature received greater changes than the risk of intensity of extreme temperature. The days of extreme temperature that occurred once in 20 years in the historical period will become once in 4 years at the 1.5°C global warming level, and will become once in 2 years at 2°C. However, the intensity of extreme temperature once in 20 years in the historical period will become once in 7 years at the 1.5°C global warming level in the future, and once in 4 years at 2℃.