As　a　nonlinear　and　time-varying　complex　dynamic　system,　wastewater　treatment　process　(WWTP)　is　difficult　to　be　controlled.　In　this　paper,　to　control　the　dissolved　oxygen　(DO)　concentration　in　a　WWTP,　a　growing　and　pruning　recurrent　fuzzy　neural　network　(GPRFNN)-based　control　system　is　proposed　which　contains　RFNN　controllers　and　RFNN　identifier.　The　identifier　is　used　to　model　the　WWTP　with　an　adaptive　algorithm　to　afford　model　information　for　the　controllers,　while　the　controllers　are　designed　to　adjust　the　control　variables　to　make　the　WWTP　run　smoothly.　Furthermore,　the　structure　of　the　RFNN　is　self-organized　to　keep　the　output　steady　in　structural　adjustment　phase,　which　is　also　theoretically　proved.　Finally,　the　control　performance　of　the　proposed　system　is　shown　by　simulation　results.