We present an application for authenticating users with mouse movements. We extract the features from nine paths between seven squares displayed consecutively on the screen. Although the paths among the squares are fixed, user can not get familiar to the paths by the selection algorithm of next path. This prevents similar results from different users. We experimented on improving the success rate, by selecting the optimal time interval between mouse movement points and refining features to obtain valid data. Experiment's results presented a false acceptance rate (FAR) of 5.9% and a false reject rate (FRR) of 5.9%.