Based on the analysis of current intrusion detection technologies, the article focuses on the application of Data Mining and Genetic Algorithm to the Intrusion Detection System. The integration of them would be competent for solving some traditional problems like the block in the course of knowledge acquisition in the expert system and the dynamic update of rules. Meanwhile, according to the practical characteristics of network intrusions, the article makes some improvements on the traditional FP growth algorithm by adopting the restrictions of the key properties to guide the process of mining, which can be profitable in discovering the frequent patterns that are more meaningful for us.