A newly developed heuristic algorithm, Harmony Search, is applied to the parameter calibration problem of the nonlinear Muskingum model. The Harmony Search could, mimicking the improvisation of music player, find better parameter values for in the nonline