With the continued increase in Internet usage, the risk of encountering online threats remains high. This study proposes a new approach for intrusion detection to produce better outcomes than similar approaches with high accuracy rates. The proposed approach uses Simulated Annealing algorithms [1] combined with Hopfield Neural network [2] for supervised learning to improve performance by increasing the correctness of true detection and reducing the error rates as a result of false detection. The proposed approach is evaluated on an intrusion detection data set called KDD99[3]. Experimental tests demonstrate the potential of the proposed approach to rapidly detect high precision and efficiency intrusion behaviors. The proposed approach offers a 99.16% accuracy rate and a 0.3% false-positive rate.