Soil erosion is one of the most important economical, social and environmental problems. On the basis of available information, Asia suffers more than the other continents from soil erosion and Iran is one of the countries that is mostly affected. Watershed management is an approach for erosion reduction. In Iran , most mountainous watersheds do not have enough hydrometery stations and defective stations have sparse data. Because of limited availability of hydrologic data, watershed modeling plays a crucial role in development of local resources. Mathematical models are employed for addressing a wide spectrum of water resources problems. In this study, SWAT2000 was applied to simulate runoff and sediment in Beheshtabad watershed, a sub-basin of Northern Karun catchment, central Iran , with an area of 3860 km 2 . This model predicts the impact of land management practices on water, sediment and agricultural chemical yields in large watersheds with varying soils, land uses and management conditions over long periods. Calibration and uncertainty analysis were performed with SUFI-2 program. The model was run for period of 1996-2004 and data of six hydrometery stations at Beheshtabad watershed were used for calibration and validation. Three-fourth of data for any station was used for calibration and the rest for validation. P-factor, d-factor, R 2 and Nash-Sutcliffe values were used to assess the ability of SWAT to estimate runoff and sediment. The results of the model to simulate monthly runoff of Beheshtabad, Kohesokhteh and Joneghan stations were satisfactory at calibration and validation stages. For calibration of monthly runoff, the p-factor, d-factor, R 2 and Nash-Sutcliffe values were 61, 0.48, 85 and 75 at Beheshtabad, 86, 1.1, 74 and 72 at Kohesokhteh and 66, 0.94, 67 and 62 at Joneghan stations, respectively. For validation of monthly runoff, these values were 53, 0.38, 85 and 57, 80, 1.33, 80 and 62 and 83, 1.01, 57 and 52, respectively. The results showed that SWAT could not simulate extreme events. For calibration of daily runoff, the model estimated runoff at Beheshtabad, Kohesokhteh and Joneghan stations satisfactorily but for validation, it only estimated daily runoff at Beheshtabad and Joneghan stations. The simulations of daily sediment were weak. For calibration of daily sediment, model estimated sediment at Beheshtabad and Kohesokhteh stations satisfactorily and for validation only estimated daily sediment at Kohesokhteh station. Fosimulation, accuracy of data and dispersed data.
