For　the　control　problem　of　dissolved　oxygen　(DO)　concentration　in　wastewater　treatment　system,　this　paper　proposes　a　modeling　and　controlling　method　based　on　feedforward　neural　network　(FNN).　For　making　neural　network　controller　control　the　uncertain　transfer　function　model,　the　approximating　neural　network　is　established　as　wastewater　treatment　model　which　using　the　approximate　character　of　FNN.　Meanwhile,　since　FNN　has　strong　online　learning　ability,　another　network　is　designed　as　controller　to　control　the　DO　concentration.　The　simulation　is　based　on　real　data　supplied　by　wastewater　treatment　plant,　and　the　results　indicate　the　DO　concentration　can　be　controlled　efficiently　with　better　control　accuracy.