Abstarct: Today, the development of more accurate methods of agricultural production is one of the strategies in order to minimize the burden of farming operations to the quality of the environment. In recent years using indirect methods such as Artificial Neural Networks and Adaptive Neuro-Fuzzy Inetrface System and other similar models to estimate these properties has been considered. Therefore a solution which predicts soil nutrient elements and also saves both costs and time, could be useful in appropriate application of chemical fertilizers in soil with nutrient deficiency. This study is founded in order to predict nitrogen, phosphorus and iron in the soil of Incheh Borun area in Golestan Province by using ANN and ANFIS models. Accordingly, twenty profiles were excavated in 5 hectare parts of each other and sampling was conducted from the depths of 0-30, 30-60, 60-90 cm. After that, the physical and chemical analysis of samples was conducted in the laboratory. Before the start of training models, training and testing data were divided into two groups that 80 percent of data used to train the model and 20 percent of which was used to test and evaluate output. then the parameters of organic matter and clay content, electrical conductivity and pH of the soil as input parameters and elements nitrogen, phosphorus and iron in the soil were as the model output parameters. that a sensitivity analysis is showed in order to determine the correlation between input and output parameters, organic matter strongest correlation with nitrogen, phosphorus and iron elements in the soil. The results of this study shows that high performance Artificial Neural Network model than Adaptive the Neuro-Fuzzy Inetrface System. In addition, the ANN has demonstrated better performance than the ANFIS in the test.