Ant　Colony　Optimization　(ACO)　had　not　a　well　efficiency　and　ignored　the　traffic　jam　which　was　a　serious　problem　in　the　real　traffic,　so　it　was　hard　to　obtain　a　good　solution.　Therefore,　a　highly　efficient　ACO　was　proposed.　First,　the　destination　emitted　fragrance　information　which　could　attract　ants　close　to　the　destination　and　make　the　search　of　ants　have　directivity;　Second,　the　route　in　road　networks　was　classified,　and　dynamic　classification　strategy　was　combined　to　avoid　the　shortcoming　such　as　stagnation.　Result　shows　that　this　algorithm　is　better　than　ACO　algorithm　in　the　stability　and　quality　of　optimum　solution.