Abstract: Based on 50-year monthly temperature and rainfall data of 48 stations in Guangdong province,the physical evaporation of Guangdong in spring is calculated with Gaoqiao equation and their eigenvectors and principal components are computed by empirical orthogonal function.The properties of temporal variation of the first four principal components are analyzed by using the methods of moving averages and climatic variability and the tendency and power spectrum.The principal component-stepwise regression model is used to predict the distribution of Guangdong water resources in spring.The results show that the mean distribution of Guangdong water resource in spring are marked by three centers of sufficient amount and two centers of insufficient amount.For the distribution of Guangdong water resource in spring,there are four basic patterns.The spring water resources of Guangdong do not have climatic variability,with the first principal component showing tendency variation by three orders of power function It is revealed that the water resources are insufficient from 1950s to early 1970s but sufficient from early 1970 s to mid-1990 s,before showing decreasing tendency again.Power spectrum analysis shows that the 1st,3rd and 4^th principal components have oscillation periods of 2.0～5.0 a.Next is a prediction of the distribution of Guangdong water resources in spring.For 2004,the type is Ⅰ_b and the distribution is serious insufficient in the whole province,and the observation agrees with the prediction exactly.For 2005,the type is Ⅲ_b,the same as the reality,but the distribution does not agree with the observation.For 2006,the type is Ⅰ_d and the distribution is the serious insufficient across the whole province,but the observation is that there is insufficient amount in the west but sufficient amount in the east.The reason why the prediction of the last two years are not successful may be related to spring barrier of sea temperature.