Aimed at the feature of annual precipitation,this paper puts forward a predicting Markov chain method based on entropy weight. Data of precipitation in Hangzhou,from 1956 to 2009,was used as an example. The precipitation can be predicted year by year using the Markov chain models based on entropy weight. Hangzhou past 5 years the results of precipitation yearly basis,respectively:the absolute error of 73mm,27mm,-22mm,-17mm and 20mm;the relative error was 5.66%, 2.03%,-1.59%,-1.08% and 1.30%.The error value of smaller than ±5% and ±10% was 36.67% and 60.00% respectively in the 30 years of precipitation prediction. M-K test was applied for nearly 30 years of predicting results for time series analysis, the results show that the prediction data with the increase in prediction accuracy tends to gradually increase.