Forecasting the runoff is an essential task in hydrological modeling. Since the factors and parameters affecting the rainfall-runoff process have spatial and temporal variations, they are engaged with significant uncertainty and complexity, leading to some errors in the modeling process, making it a very unwieldy task. This study uses the genetic programming (GP) model to simulate the rainfall-runoff process in Khorramabad River basin located in Iran. To validate the performance of the proposed GP method and accomplish this kind of modeling, this technique is compared to the HEC-HMS conceptual model. For this purpose, the HEC-HMS model is calibrated and employed to simulate the rainfall-runoff process at the first step, modeling four flood events in the study area. In the second step, the GP Model is run as a rainfall-runoff model in the mentioned area. Then, the performance criteria, including the root mean square error (RMSE) of the peak discharge, mean absolute error (MAE), and the observed and simulated flow volume difference, are used to compare two simulation approaches. For example, in the simulation stage of the year 2014, the values of the two mentioned criteria for the genetic programming model were 1.22 and 0.99, while for the hack model, the values were 7.28 and 9.75, respectively. The results suggested that the GP as a data-driven model performs better than the HEC-HMS model as a physics-based one to simulate the rainfall-runoff process in this basin.