Since the grounding mode, line parameters, and thus the transient performance of MMC-HVDC are quite different from those of LCC-HVDC system and AC system, fault detection strategies used in LCC-HVDC are no longer applicable to MMC-HVDC. In this paper, a wavelet entropy based fault detection approach for MMC-HVDC is proposed to detect DC line faults and distinguish AC area fault. This method does not use the magnitude of the current but evaluate the wavelet entropy of the current signal to detect the fault. Firstly, the factors which affect fault transient performance of MMC-HVDC are analyzed. Then the wavelet entropy theory and its role in the detection of non-stationary signals are outlined. The proposed fault detection mechanism using wavelet energy entropy and wavelet time entropy to process fault current signals is described in detail. Finally, the computer simulations are carried out. The results show that the proposed method can detect the pole to ground fault and the pole to pole fault in MMC-HVDC accurately and can satisfy the time requirement.