Among the processing conditions of injection molding, temperature of the melt entering the mold plays a significant role in determining the quality of molded parts. In our previous research, a neural network was developed to predict the melt temperature in the barrel during the plastication phase. In this paper, a neural network is proposed to predict the melt temperature at the nozzle exit during the injection phase. A typical two-layer neural network with back propagation learning rules is used to model the relationship between input and output in the injection phase. The preliminary results show that the network works well and may be used for on-line optimization and control of injection molding processes.