With the rapid development of information technology, computer networks are exposed to a wide variety of vulnerabilities and an increasing number of security threats. Intrusive traffics affects the normal functionality of the network's operation. Thus, a need for more intelligent and sophisticated security controls such as intrusion detection systems (IDSs) is necessary. Since IDS has to deal with problems such as large network tra c volumes and di culty to realize decision boundaries between normal and abnormal behaviors, classification plays an important role in the detection. Machine learning approaches have been extensively used in network intrusion detection techniques because they require less human expert knowledge, significantly reduce the burden of analyzing huge volumes of network traffic, and provide more precise results by separating data into different classes (normal and abnormal) as correctly as possible with the help of a model. The experimental results indicate that using significant features instead of all features improved the performance of classification techniques based-IDS, and hence shows significant improvement in detecting the intrusions correctly.