A well developed network of roads must exist as a necessary infrastructure system in modern forestry to facilitate forest operations. But forest roads have the potential to disrupt the drainage characteristics of watersheds and lead to negative impacts on the environment with increased erosion and sediment yields. Numerous factors affect surface erosion of roads and sediment production potential; determining and ranking them could be a guide for management decisions to erosion control. In this study, the CULSED model (as an extension of ArcGIS) was used to estimate sediment delivery and the distribution of a road network, given the existing culverts. Using the model, some culverts were added to the road network around places with high sediment delivery in order to minimize it. After a correlation analysis and adjustment between sediment production and the factors, i.e., road width, road gradient, age of road and vegetation cover, the trend of changes in sediment delivery with model changes in the input was investigated with a sensitivity analysis of the model. The results show that adding new culverts to the road resulted in a significant reduction of sediment delivery. The most important factor affecting sediment delivery was road width, followed by road gradient, vegetation cover and age of road. Road width and gradient were positively correlated with sediment delivery, while vegetation cover and age of road were negatively correlated. The best model to show the relation between sediment delivery and road width as well as with road gradient was a linear model, for vegetation cover a cubic equation and for road age a power model. The results of sensitivity analysis showed that sediment delivery had the greatest sensitivity to changes of road width and was least sensitive to changes in the age of the road. This model can help to estimate sediment delivery with its spatial distribution, which can be used for optimization of cross drain systems and strategies of sediment control. Application of the model requires field trials to acquire the necessary input data. The reliability of our results is a function of the accuracy of inputs, especially digital elevation model. A well developed network of roads must exist as a necessary infrastructure system in modern forestry to facilitate forest operations. But forest roads have the potential to disrupt the drainage characteristics of watersheds and lead to negative impacts on the environment with increased erosion and sediment yields. Numerous factors affecting surface erosion of roads and sediment production potential; determining and rating them could be a guide for management decisions to erosion control. In this study, the CULSED model (as an extension of ArcGIS) was used to estimate sediment delivery and the distribution of a road network, given the existing culverts. given the existing culverts were added to the road network around places with high sediment delivery in order to minimize it. After a correlation analysis and adjustment between sediment production and the factors, ie, road width , road gradient, age of road and vegetation cover, the trend of changes in sediment delivery with model cha nges in the input was investigated with a sensitivity analysis of the model. The results show that adding new culverts to the road resulted in a significant reduction of sediment delivery. The most important factor affecting sediment delivery was road width, followed by road gradient, vegetation cover and age of road. Road width and gradient were positively correlated with sediment delivery, while vegetation cover and age of road were negatively correlated. The best model to show the relation between sediment delivery and road width as well as with road gradient was a linear model, for vegetation cover a cubic equation and for road age a power model. The results of sensitivity analysis showed that sediment delivery had the greatest sensitivity to changes of road width and least least sensitive to changes in the age of the road. This model can help to estimate sediment delivery with its spatial distribution, which can be used for optimization of cross drain systems and strategies of sediment control. Application of the model requires field trials to acquire the necessary input data. The reliability of our results is a function of the accuracy of inputs, especially digital elevation model.