This suggest the prices forecast model for international grains using artificial neural network. For this, we first select the most suitable network models for grain prices and compare the predictive power with ARIMA (autoregressive and moving average model) and GARCH (generalized autoregressive conditional heteroskedasticity) models. The results show that the predictive power of artificial neural network is similar to or lower than the alternative models. When only past prices are considered, the neural network model has the higher predictive power compared to the cases with other market variables. Implications for forecasting of international grain prices are suggested at the end of this study.