Abstract: In order to retain the advantage of gait recognition and to overcome the shortcomings of single biometric and increase recognition rate for human recognition at a distance, a novel approach of fusing gait and side face for human recognition at the feature level was presented. Firstly, the feature exaction and dimension reducing was done to Gait Energy Image (GEI) and Side Face Image, respectively, by Two-Direction Image Matrix based Principal Component Analysis(2DIMPCA), and two original feature vectors were obtained correspondingly, which are vectored and integrated into synthetic feature vectors. Then Multiple Discriminant Analysis (MDA) was employed on the synthetic feature vectors of gait and side face to obtain fusion features vectors. Finally, the recognition process was implemented on the fusion feature vectors by nearest neighbor (NN) algorithm. Experiments were implemented on Dataset B of CASIA gait databases and the results showed that the higher correct recognition rate was gained. In the meantime, the validity of the method was demonstrated, and a new approach was supplied for multimodal biometric identification.