Roving patrol plays a very important role in strengthening safety within campuses or communities through their patrol around the target. In fact, there maybe exist some patrolmen who do not work efficiently. Therefore, it is useful to quantitatively evaluate their roving effectiveness. The widespread use of global positioning system (GPS) and GPS data logger have generated huge amount of trajectory data. It is a good way to evaluate the effectiveness. In this paper, we present the roving patrol effectiveness evaluation (RPEE) model by GPS Trajectory. Firstly, GPS data is pre-processed by redundancy reducing, trajectory segmentation and abnormality filtering. And then, evaluation index system is established based on the indicators extracted from trajectories by using statistical analysis. Finally, RPEE is conducted by using fuzzy comprehensive evaluation method. In this procedure, the weight distribution for indicators is determined by analytic hierarchy process. We evaluate the model by using the GPS data collected at 11 distinct patrol areas over a period of six months in the real world. The experimental results show that: indicators, such as the total effective patrol time, the phenomenon of staying, and the roving patrols at each checkpoint, etc., can be obtained objectively for RPEE.