The least mean square adaptive filter (LMS-AF) and the variable step size LMS-AF (VSS-LMS-AF) have been widely used to suppress the motion artifact (MA) from ambulatory electrocardiograms (AECG).How- ever,due to the fact that the non-stationary property of the MA and the AECG increases the mis-adjustment of fil- ters,although the combined AF (C-AF) has advantages in tracking non-stationary signals，it still suffers from a signal distortion.In order to further decrease the signal distortion and improve the MA suppression ability,a new feed-forward combined AF (FFC-AF) is proposed,in which the stationary degree of the MA is estimated by using a reference signal and two parallel AF outputs are combined.Then,a simulation experiment with the AECG re- cording on ten healthy volunteers is conducted to assess the performance of the FFC-AF.The results show that,as compared with the C-AF,the FFC-AF is of less signal distortion and larger signal artifact ratio.Therefore,it can be used to suppress the MA in the AECG.