Abstract: Take Jinan karstic spring as an example, the statistical method combined with the statistical software SPSS is used to analyze the regime of karst aquifer system. According to the 38 years＇ monitoring data, spring flow regime process was divided into two periods （1959-1968 ＆ 1969-1996）, and partial correlations analysis has been made in two different period. The result shows that natural factor is main factor in the previous period. However, natural factor and human activity are the main factor in the latter period. On the basis of partial correlation analysis, mathematical model of each period is developed and the spring flux is simulated and predicted. The error of simulation period is 2% and prediction period （1981-1987） is 3.4%. The fitting and prediction result is good.