The performance of artificial neural networks on building process models for estimating the quality of petroleum products, research octane number of gasoline, boiling point of gas oil of a topping unit, and flash point of bottom product of a naphtha splitter, are examined in this study. Three types of artificial neural networks models are developed in this study; back propagation neural network, radial basis function neural network and Wave-net. It is shown that radial basis function neural network model and back propagation neural network model are superior to the other neural networks models on building a steady state model, Wave-net is useful in building a dynamic model for time series data.