Power quality is increasingly becoming a major concern due to excessive use of sensitive equipments. Determining the cause of power quality disturbances using various approaches such as neural network and expert system becomes essential in order to take suitable corrective action. In this paper, wavelet transform is proposed to identify the predominant harmonic content, the instant of occurrence and take suitable corrective action. Comparison between mean absolute deviation (MAD) levels for a pure sinusoidal and polluted waveform is used to detect the predominant harmonic content. The system has been simulated in MATLAB and Simulink environment. © 2005 IEEE.