Abstract: In order to improve the prediction accuracy of additional crack percentage for paddy rice drying in a deep fixed-bed, Adaptive-Network-based Fuzzy Inference System (ANFIS) was applied to establish a prediction model for the additional crack percentage. Through verification of the prediction model, it was determined that the maximum prediction error was 14.57%, the minimum prediction error was 1.68%, the average prediction error was 5.68% and the prediction accuracy reached 94.32%. The results of analysis show that the prediction accuracy and the generalization of the prediction model are very high, and the model can predict the effects of drying parameters on additional crack percentage conveniently, which contributed to understanding of variation of additional crack percentage impacted by drying parameters accurately and provided the foundation for selecting drying parameters properly and for controlling drying quality.