Abstract. This paper discusses concurrency control in distributed geographical database systems. A geographic data server providing transactional services must be designed to handle both long- and short-duration transactions. The main challenges are identified as: (1) preserving the isolation property when concurrent transactions perform read and write operations on pages; (2) ensuring that search and other operations on the spatial access structure appear isolated; and (3) whatever method that is chosen they must provide feasible solutions with respect to performance. Existing approaches are discussed and presented.