In this paper an efficient optimization technique called Harmony Search (HS) algorithm is considered for IIR adaptive system identification problem. The proposed technique HS performs a structured randomized search of an unknown vector within a multidimensional search space for finding the optimal solution. Better exploration and exploitation of entire search space ensure the avoidance of premature convergence and stagnation problems which are more or less usually present with conventional RGA, PSO and DE. The simulation results justify the optimization efficacy of the proposed HS based identification technique over RGA, PSO and DE.