To protect the spatial vector data, which is often in an irregular shape and distributed throughout multiple sensitive areas, the traditional mandatory access control model is extended and explained in this paper. This paper also proposes a fine-grained spatial mandatory query access control model?SV_MAC (spatial vector data mandatory access control model). Also, an AR+ spatial index tree technique is advanced, which combines the search of both spatial data and access control policies together to efficiently enforce the SV_MAC model in the course of spatial vector data searching. Experiment results shows that AR+ tree can not only provide fine-grained security protection for sensitive spatial vector data, but can also guarantee good user experience for GIS (geography information system) applications.