in order to investigate the efficiency and accuracy of the gene expression programming method in predicting the suspended load sedimentation of the Navroud River in Gilan province using the software (GEP), from the data of average rainfall, average runoff discharge, average maximum temperature , the average minimum temperature and the average annual temperature of the basin, as input variables, as well as the observational suspended load sediment data at the Kharggil Aslam hydrometric station, as the output variable in the joint statistical period of 20 years (from the water year 74-75 until 93-94) was used. For this purpose, the amount of sediments of the suspended load of the basin using the gauge curve method, the average amount of rainfall using the method of precipitation curves using the Kriging model in the GIS software environment and the values of three annual temperature indices through the relations of the thermal gradient of the region for the statistical years It was estimated differently. In order to achieve the appropriate structure between the inputs and the output of the model, the information of the first 13 statistical years was used to train the program and the remaining 7 years were used for its verification, and in order to identify the effective inputs, 6 patterns were used. Various were designed based on the combination of input data. Finally, to determine the most suitable relationship for predicting basin sediments, two valid indicators of root mean square error (RMSE) and correlation coefficient (2R) were used. The results showed that, with the structure of average rainfall, average runoff, average maximum and average annual temperature of the basin, the correlation coefficient (2R) between the calculated and observed data in the verification period reached its highest value (84%) And the value of the RMSE index reaches its lowest value (1.06).