In this paper we present an efficient data structure, called G^r_tree, for organizing multidimensional data. The proposed structure combines the features of distance functions of metric spaces and G-trees. Although the G^r_tree requires distance computations and has the overhead of a small amount of storage space, due to the introduction of active regions inside the partitions of the data space, it reduces the accesses of partial match and range queries. We give algorithms for the dynamic operations of the G^r_tree, examine several types of queries and provide some comparative results.