The average value of effective accumulated high temperature in summer for many years was considered as a standard to judge the extent of a hot summer. The CART algorithm was employed to explore the relationship between the high temperature and the climatic factors such as East Asian summer monsoon index, summer India-Myanmar trough, summer North Atlantic Oscillation (NAO) and equatorial Pacific sea temperature, and the high-temperature prediction rule set was obtained to build a high-temperature prediction model. The study selected the daily maximum temperature data in summer among 1955—2012 in Zhangzhou of Fujian Province. The extent of a hot summer was determined through the effective accumulated temperature of high temperature in summer for 58 years. A number of climatic factors in the same period were input as the input variables, and 46a data were randomly selected to get 10 classification rule sets. The accuracy of the built predication model reached 91.49%. The remaining 12a data were used for test, with an accuracy up to 91.67%. Generally speaking, the results of this paper have verified the feasibility and validity of the high temperature prediction model, which provides a new idea for the research of the catastrophic weather model.