Gait recognition is an emergent technique in human identification that has been used widely in surveillance and forensic applications. This paper investigates the recognition of gait using motion capture data. Normalization method is applied to constant the data. Principle Component Analysis (PCA) is used to reduce the dimensionality of the constant gait motion data. The principle components obtained were then being matched using Euclidean Distance method between the sample and subject data. In our project, we recognize the action of person using the GAIT samples of a person. Here, we extract gait information of a person from the given video. With the help of gait information we track the person in the video. Then by using PCA we can find the action of silhouette sample sequences for a person. Finally the action of person for the given video was recognized.