There is no doubt that climatic factors are one of significant parameters in occurrence of natural fires in forest and range ecosystems. The goal of this study was a monthly-based prediction of the occurrence of the natural fires using artificial neural networks in Izeh, north-west of Khuzestan province. The natural fire occurrence data including date of the occurrence, the burned area and number of the fire occurrence was obtained from Izeh Natural Resources Office. The findings indicated that the algorithm of multiple layer perceptron and hyperbolic function were efficient in exploring the relationship between climatic factors and the natural fire occurrence. The networks with two hidden layers and 15 neurons have revealed high accuracy in prediction of the natural fires occurrence. Moreover, for prediction step FMSE(Final Mean Square) was recorded 0.0038. While for testing step, coefficient of variation, MSE(Mean Square), and NMSE(Normal Mean Square) were equal to 0.99, 0.073, and 0.018, respectively. For validation step, the trained network has indicated a high determination coefficient (r2=0.98) between the observed and predicted values. It should be mentioned that the present approach in this study could achieve an artificial neural network with medium performance (r2=0.58) between climate data and the burned area of the natural fire.