Abstract: Although BP algorithm is a very important method in the intrusion detection systems (IDS), it can cause the system problems of long time detection and low detection rate. To solve these problems, through analyzing the BP algorithm, a novel EBP algorithm (Enhanced BP algorithm) for IDS is constructed. An improved error function is designed in EBP algorithm to solve some disadvantages of the IDS, such as the slower detection speed and detection efficiency. Moreover, a self-adaptive method of magnifying error signal is also proposed. The results show that the EBP algorithm is an efficient algorithm. Compared with the classical BP algorithm and the MBP algorithm, the EBP algorithm can not only increases the convergence speed of network training obviously, but also improves the detection rate of IDS to 14.0% and 8.1% respectively. Besides, the decrease in false positive rate of EBP algorithm is 4.45% and 1.5% than other two algorithms respectively.