Prediction of destructive flood events particularly in degraded watersheds necessitates the importance of model calibration procedure. Multi-objective calibration of hydrologic model parameters with conflicting objectives attempts to adjust the parameter values in terms of different objective functions. Thus, this research carried out a procedure of multi-objective optimization for a distributed and single event-based rainfall–runoff model (i.e. Kineros2) through the AMALGAM algorithm in MATLAB environment. Four rainfall events with different durations and intensities in Tamar watershed, located in the northeast of Iran, were considered for rainfall-runoff simulation. Three objective functions including Nash- Sutcliffe Efficiency (NSE), Error in the Stage of Peak flood (ESP) and Relative Time Shift (RTS) were simultaneously employed during optimization process. The model optimization was evaluated through 5000 epochs and the best values of NSE (0.91), ESP (0) and RTS (0) were obtained for the simulation based on the third storm event (05/10/2005). The non-dominated solutions, extracted from the AMALGAM approach were plotted in a three-dimensional space to form the Pareto front. According to the inspection of interactive regions on the Pareto fronts, it was discovered that the behavior of RTS and ESP functions was in the same direction. Therefore, it should be expected that, the similar results will be achieved even by employing merely NSE and RTS as the fitness functions, especially in the simulations involved with flood warning systems.