Due to the time and spatial limitations of subsurface drainage pilots, simulation models have been extensively applied for evaluating these systems. Since the accuracy of simulation models depends enormously on the accuracy of model parameters, this study aims to develop an inverse modeling approach for estimating most influential soil hydraulic and solute transport parameters in a subsurface drainage system in an arid and semi-arid region. The SWAP model in conjunction with a genetic algorithm and PEST optimization tool was used to find optimum parameters by minimizing the differences between observed and simulated values of drainage discharge, watertable depth, and drainage salinity. Results revealed that the best simulation of drainage outputs was obtained by parameters which were estimated minimizing an objective function that included all three datasets via a genetic algorithm. Although assuming the soil as a homogeneous and heterogeneous medium had quite similar results from objective functions with one or two datasets, homogeneous assumption worked better in the objective function with three datasets. The inverse modelling approach with GA resulted in a better performance as compared to the PEST optimization tool, particularly in objective functions with two or three datasets.