Power quality (PQ) analysis is the foundation of power system automation. The premise of power quality analysis is feature representation of power quality events. Time-frequency analysis (TFA) is very suitable for nonstationary signals analysis. The TFA of a PQ signal is to determine the energy distribution along the frequency axis at each time instant. This paper provides a status report of feature representation for PQ events by TFA methods, including short time Fourier transform (STFT), wavelet transform (WT) and S-transform (ST), overview the basic TFA theories for PQ analysis and compare the effectiveness of different TFA methodology. The expectation is that further research and applications of these TFA algorithms will flourish for PQ feature representation in the near future. The analysis direction and emphasis of studying are also put forward.