The physical properties of pistachio kernel are necessary for the proper design of equipments for transporting, drying, processing, sorting, grading and storing this crop. In this study, physical properties of pistachio were simulated via different models of neural networks. Different models of neural networks with different trishold functions were used to forecast surface area, volume, mass and kernel density of pistachio. The results showed that radial basis function neural network with ordinary radial basis function trishold function had a favorable results to forecast surface area, volume, mass and kernel density of pistachio and this network could predict surface area, volume, mass and kernel density with R2 value 0.982, 0.982, 0.992 and 0.962, respectively. Furthermore, in this research surface area, volume, mass and kernel density of pistachio were fitted by regression equation, the result showed linear regression method could predict surface area, volume, mass and kernel density with R2 value 0.931, 0.897, 0.985 and 0.944, respectively. Generally, the result showed neural network model had a higher ability to forecast physical propertied of pistachio than linear regression method.