IPv6 technology introduced to solve problems in the previous IPv4 technology. New threats were discovered due to the exploitation of flaws in IPv6 architecture design. There is a demand to produce new intrusion detection technique for new threats of IPv6 network environment. In this paper, the method of features selection to produce the most significant feature is presented. The objective of this paper is to propose a framework to solve feature selection problem which in this paper the features of IPv6 packet will be the case. The fundamental method of feature selection was improvised to suit with this scenario. A data of IPv6 network attacks was produced by using an IPv6 testbed environment. The SVM and PSO were used in the process of determine the best features to detect IPv6 attacks. In the future, this framework can be applied in other domains which require features selection solution.