According to reports of forest range and watershed management organization of Iran, hundreds of forest fires take place in our country every year. Due to destroying forest and grass land in different parts of Iran, especially in Zagross and Alborz Mountain, it is necessary to develop new methods to reduce fire effects in these places. There are many parameters that increase probability of fire occurrence. Without considering these parameters, environmental and biomass impacts are very critical problems. As fire could be occur in many grass lands and forests, fire monitoring of fire in vegetation regions is impossible. Using Remote Sensing technology and GIS modeling is a basic way to monitor and prevent from this national disaster. In this study, spatial distribution of fires which occurred in the last 8 years in Iran and were detected by MODIS satellite images was used. Then, by using NDVI index and considering repetitive fires, permanent fires were distinguished from random fires. Finally, density function and spatial autocorrelation in GIS were employed as statistical parameters to detect potential regions to fire.