Gait-based human identification aims to discriminate individuals by the way they walk. A unique advantage of gait as a biometric is that it requires no subject contact and is easily acquired at a distance, which stands in contrast to other biometric techniques involving face, fingerprints, iris, etc. This paper proposes a new gait representation called motion energy image (MEl). Compared with other gait features, MEl is more robust against noise that can be included in binary gait silhouette images due to various factors. The effectiveness of the proposed method for gait recognition is demonstrated using experiments performed on the NLPR database.