Under　the　application　background　of　network　security　evaluation　research,　this　paper　proposes　a　method　of　situation　prediction　based　on　Particle　Swarm　Optimization　(PSO)　for　optimizing　BP　Neural　Network　(BPNN).　It　uses　PSO　to　reach　global　optimization　of　BP　network＇s　weight　value　and　threshold　value,　and　then　by　means　of　the　optimized　BP　network　builds　a　prediction　model　to　predict　the　future　network　security　situation.　Experiment　results　show　that　this　method　can　overcome　the　shortage　of　the　predicting　application　in　the　traditional　BP　network,　and　effectively　improve　the　accuracy　of　situation　prediction.　It　can　be　applied　into　the　situation　prediction　of　network　security　situation　awareness.　©　2008　IEEE.