Abstract: A multi-status Markov chain model is proposed to simulate daily rainfalls records and the extreme rainfalls in summer are fitted from the simulated daily rainfall records by using a Generalization Pareto Distribution (GPD) model. The research results show that this statistical simulation method display most good dependability and the statistic climatic features from the simulated daily rainfall records have reached more high precision for most stations, specially the pluvial region over east area of China. The analysis comparatively shows that the multi-status Markov chain model is excellent with the two status Markov chain model for its simulation ability, specially, the simulated climatic features of extreme rainfalls in east region of China have reached most high precision. The experiment results for the selected six stations demonstrate the excellent simulation effects of above statistical model, for example, the following features: statandard deviation of monthly rainfall, maxmum daily rainfall, mean number of rain days during a month spell, statandard deviation of daily rainfall, mean value of daily rainfall amount etc. are consistent with the real obvervation values. Both the multi status Markov chain model and the two status Markov chain model.