Since, floods have a devastating effect on human life and their economy, the use of modeling to predict the flow of rivers is very essential and makes it possible to quantify the effects of flows on a certain area. The aim of this research was to compare discharge prediction efficiency of Soil and Water Assessment Tool (SWAT), Hydrological Engineering Center-Hydrological modeling System (HEC-HMS), Hydrologiska Byråns Vattenbalansavdelning (HBV) and Parameter Efficient Distribution (PED) models in Borkena watershed for future time stream flow prediction. The study area is found in the upper part of Awash basin which covers 1669 km2. According to the sensitivity analysis results, average slope length, deep aquifer percolation fraction, threshold depth of water "revap", effective hydraulic conductivity in channel alluvium were very sensitive parameters for SWAT and recession constant, initial abstraction and curve number are also sensitive parameters for HEC-HMS model. For HBV models, upper zone of the soil layer, field capacity routing length of weighting function and for PED hill side area are the most sensitive parameters. The models were calibrated from 1999 to 2009 and validated from 2010 to 2015. On the daily time step flow validation results, for SWAT model NSC (0.681), R2 (0.688), PBIAS (-6.5) and RMSE (14.238), for HEC-HMS model NSC (0.66), R2 (0.664), PBIAS (0.6) and RMSE (17.45), for HBV model NSC (0.653), R2 (0.73), PBIAS (27.34) and RMSE (17.63), for PED model NSC (0.654), R2 (0.704), PBIAS (10.28) and RMSE (0.907). Based on the validation results, SWAT model perform well based on the common performance indicators which are NSC, R2 and PBIAS results than the other models. According to the results, when comparing models in discharge estimation efficiciency (Nash Sutcliff Efficiency), SWAT model was the first and the second one was HEC-HMS model. The third and fourth models were HBV and PED respectively. The overall results indicate that the two infiltration excess (SWAT and HEC-HMS) models perform in a better way than the two saturation excess (HBV and PED) models.