Biometric data are required for government surveillance to maintain homeland security. The common biometric methods include fingerprints, retinal patterns, and signatures. Unfortunately, those methods encounter several pitfalls, such as inquiry for subject's consciousness, heavy computational resources, and falsification risk. We propose gait analysis as a complementary complement for biometric surveillance approaches due to its accuracy in distinguishing different persons. Using a Kinect™ device, gait analyses are expected to be low-cost and easy to use. During the experiment 2 male subjects in the age of 24-26 were involved. One of them was in the normal weight category, while the other was overweight. The subjects were asked to walk on a 2.4-m aisle towards the device 5 times. Next, the recorded images are interpreted to obtain the joint positions of each subject using a Dump Kinect™ Skeleton software. The results are presented in the Cartesian coordinate (x, y, z) to represent the position of each joint per frame. However, we were only focused on the 5 joints in the axial skeleton. The analysis of variance (Anova) results within a subject showed that the movements of base spine, middle spine, and shoulder spine were similar, indicating the validity of the Kinect™. Subsequently, the movements of the 3 joints from Subject 1 and Subject 2 were compared. The t-test results showed that the gait of Subject 1 and Subject 2 were significantly different. It means that the device can distinguish 2 different persons through their gait. It also confirms the previous studies. Furthermore, these results also indicated that a Kinect™ device can be further developed into gait-based surveillance tool.