In data grid, large quantity of data files is produced and data replication is applied to reduce data access time. Efficiently utilizing grid resources becomes important research issue since available resources in grid are limited while large number of workloads and large size of data files are produced. Dynamic replication in data grid aims to reduce data access time and to utilize network and storage resources efficiently. This paper proposes a novel dynamic replication strategy, called BHR, which reduces data access time by avoiding network congestions in a data grid network. With BHR strategy, we can take benefits from ‘network-level locality’ which represents that required file is located in the site which has broad bandwidth to the site of job execution. We evaluate BHR strategy by implementing it in an OptorSim, a data grid simulator initially developed by European Data Grid Projects. The simulation results show that BHR strategy can outperform other optimization techniques in terms of data access time when hierarchy of bandwidth appears in Internet. BHR extends current site-level replica optimization study to the net- work-level.