Abstarct: Estimation of suspended sediment load is an important factor in water engineering, hydraulic and environment. Several attempts have been made to determine the suspended sediment load accurately. Most of these attempts have been tried to relate the amount of sediment discharge to flow discharge for estimating the suspended sediment load. But these conventional regression methods have uncertainty. So, a number of researchers have been attempted to estimate the suspended sediment load by using intelligence methods such as artificial neural networks, adaptive neural fuzzy inference system and evolutionary algorithms. In this research, Gene exxpression Programming (GEP) has been developed to predict the suspended sediment load of Jelogir and Payepol stations located in the upstream of Karkheh reservoir dam. The results have been compared to sediment rating curve and FAO methods results. To achieve this aim, daily stream flow, suspended sediment discharge and rain height corresponding from 1365-1390 were collected and used. So, 2 scenarios were defined for GEP performing. The first one has been performed by using daily stream flow and suspended sediment discharge data. The second one used the rain height information of catchment, too. The results showed that the second scenario was more accurate rather than the first one. The comparison results of test period indicated that the GEP has decreased the root mean square error (RMSE) and mean absolute error (MAE) around 91% and 94% for Jelogir station and 60% and 71% for Payepol station in relation to USBR results. Also, the results of test period denoted that GEP has decreased the RMSE and MAE around 92% and 96% for Jelogir station and 85% and 95% for Payepol station compared by FAO results.