Abstarct: In this study, using the Artificial Neural network (ANN) and adaptive-fuzzy inductive inference system (ANFIS) have used to predict some soil micro elements. The study was carried out in the city of Cherram and Dehdasht with 90 surface soil samples. For this purpose, easily found soil parameters such as soil lime, pH, EC and organic matter, iron and zinc were used as input data to the model. The results revealed that correlation between iron and zinc elements with input parameters showed the highest correlation between organic matter and clay content with the iron element, acidity and electrical conductivity with soil elements. The results of perceptron network (MLP) for calculating iron and zinc elements showed that the best explanation coefficient was compared with other functions related to Gaussian function, which was 0.93 for sigmoid tangent and 0.91 for sigmoid logarithm, and 0.31 for the sigmoid tangent, 0.3 for the sigmoid logarithm for iron and zinc, respectively. The results of the implementation of the Fuzzy Adaptive Inference System model for prediction of iron and zinc elements illustrated that ANFIS model with Gaussian membership function for both iron and zinc elements and having a coefficient of 0.99 for iron as well as a coefficient of explanation of 0.78 for zinc has the best mode over other functions. The results of the comparison of the two models indicate that the ANFIS model has a coefficient of 0.99 and 0.78 for iron and zinc elements, respectively, compared with the artificial neural network model with a coefficient explanation of 0.93 and 0.91, and also, 0.31 and 0.30 are the best in iron and zinc elements, respectively, in the training and certification phases. The ANFIS model's explanation coefficient is higher than the neural network, and the error rate of the test, the test error and the simulation error for iron and zinc elements in the neural network are higher, which indicates the lower efficiency of the artificial neural network model compared to ANFIS model.