Hydantoin epoxy resin, with a hydantoin group, is a kind of low viscosity nitrigen-containing epoxy resin. We prepared methyl hexahydrophalic anhydride (MeHHPY)/ hydantin epoxy resin, in which MeHHPA was used as cure agent. Non-isothermal differential scanning calorimetry was examined to follow the curing process. The variation of viscosity was measured the isothermal curing process by rotating viscometer respectively (70-90°C). A viscosity model, which parameters were determined by Arrhenius equation, was established on the basis of experimental data at three different temperatures. As the result showed, there is a minimum deviation comparing the dual Arrhenius viscosity model data and experimental data in the temperature range of this research. The potential of artificial neural network techniques (ANN) was employed to analyze and predict the chemorheological behavior of MeHHPA/hydatoin epoxy resin. A three layer feed forward ANN model having two input neurons, one output neuron and fourteen hidden neurons was developed to predict the chemorheologica behaviour of MeHHPA/hydatoin epoxy resin. The learning of ANN was accomplished by a backpropagation algorithm. The results display that prediction model has very good accuracy in the process of the whole experiment. Through the established ANN model, the variation characteristic of viscosity can be exactly predicted. Studying on chemorheology by ANN model can help to formulate and optimize technical process.