<p>Generally, the flow of network packets contains a huge number of attributes. Therefore, it is crucial to eliminate irrelevant features to model an efficient Network Intrusion Detection System (NIDS). In this study, a novel approach called Feature Fusion based Weighted Multi-Objective Grey Wolf Optimizer (FF-WMOGWO) is proposed for selecting the important traffic features to build an efficient NIDS. In this proposed approach, filter-based feature selection strategies, including chi-square, mutual information, feature importance, and minimum redundancy, are fused to remove irrelevant features and categorize the remaining features based on weights determined by the feature fusion. Then, Multi-Objective Grey Wolf Optimization with weighted population initialization is applied to enhance the exploration in the NIDS feature search space and to improve the efficiency of the NIDS. The proposed FF-WMOGWO is evaluated on the Canadian Institute for Cybersecurity Intrusion Detection System (CIC-IDS) 2017 dataset and improved Knowledge Discovery in Database (KDD) using various machine learning techniques like Decision Tree, K-Nearest-Neighbour, Naive Bayes, Ada Boost, Support Vector Machine, and its performance is compared with other cutting-edge NIDS. The assessment results show that the proposed approach with a decision tree classifier produced the average recall of 95.8%, precision of 97.2%, F1-score of 95.6% and accuracy of 98% on the improved NSL-KDD dataset. Also, it produced the average recall of 96%, precision of 98.9%, F1-score of 97.3% and accuracy of 100% on CIC-IDS2017. The proposed method performs better than other techniques and achieves a significant attack detection rate even for attacks with a lower sample. So, the proposed FF-WMOGWO can provide an effective and efficient approach for feature selection for NIDS.</p>