In applied hydrology, predicting peak flow for a stream or river is so complex due to temporal and spatial dependency of hydrological variables such as meteorological parameters, variations in soil type and land use. Either advanced distributed hydrological models or simple Lump models can be used for simulating these situations. This paper compares the performance of the quasi-distributed model ModClark versus lumped parameter model Clark in simulating the process of transformation of rainfall to runoff. The aim of this comparison is to identify whether using a complex model which takes into account spatial and temporal distribution parameters, which are hard to prepare and use, will lead to more precise results or not. For the purpose of this study, historical data of Randan basin situated in semi-arid region of Iran in North West of Tehran was used. The size of the catchment is 67.76 km 2 . Reviewing the results of calibration and accuracy of models revealed that both models are able to simulate the hydrology of the catchment in an acceptable way.