Real-time traffic flow prediction is one of important contents of intelligent transportation system research. Combined with the related knowledge of wavelet analysis and fuzzy neural networks, this paper gives the traffic flow forecasting model based on wavelet fuzzy neural networks. It takes wavelet function as fuzzy membership function, uses neural networks to realize fuzzy reasoning, and finishes the estimation of next cyclical traffic flow. Simultaneously the genetic algorithm is used to optimize the overall network. After the field data test, this method has high forecasting precision, stable convergence process, strong adaptability.