mras (model reference adaptive system) to achieve frequency domain performance specifications with different transfer function identification techniques are experimentally studied. standard open loop and closed loop recursive least squares (rls) system identification techniques are implemented along with methods based on bandpass filters and steady-state kalman filter. the behavior of the mras when combined with the different system identification techniques is discussed. based on experimental work performed on mras using open loop system identification, it was found that the system can achieve performance specifications in all process conditions. it was noticed that the dc bias changes in process input when process changes, delays the convergence of the system identification in open loop system identification. closed loop system identification was found inadequate for use in mras. the system identification based on bandpass filter approach was found slow in convergence and so inefficient for use in mras. the use of kalman filter for system identification resulted in noisy gains on adaptive pi controller. based on the results obtained, it is concluded that mras to achieve frequency domain performance specifications is practical and capable of being used in industrial process control when coupled with open loop identification based on frequency domain concepts. the proposed approach provides a better method for maintaining performance robustness while guaranteeing stability margins in adaptive control than that obtained from time domain adaptive control.