Abstract: Using the 19 global climate models, ensemble models in conjunction with the observational data in Liaoning province, we evaluated the accuracy and credibility of air temperature simulated by different models under different typical concentration paths.The results indicate that the optimal model simulation performs better and has higher credibility than ensemble simulation.With the increase in emissions and concentrations of carbon dioxide, the change rate and credibility in simulated air temperature tend to show an increasing trend, and the beginning years when the temperature increment is beyond 2 ℃ for the first time, occur before 2011 at most stations.The later the beginning year is, the higher the temperature increments, and vice versa.The beginning years with temperature increment of stably by 2 ℃ occur before 2022 at most stations, and the ending years occur between 2019 and 2016, with a duration of less than 13 years.The beginning years occur earlier in the western region than in the eastern region in Liaoning province, while the ending years and duration time distribute more even.With the increase in global emissions and concentrations of carbon dioxide, the warming rate increases.Occurrence time and variation trend of the maximum, minimum, and average temperatures in different regions under different typical concentration paths are basically consistent with each other.