To account for spatial variability of precipitation, as well as basin physiographic properties, the National Weather Service (NWS) has developed a distributed version of its hydrologic component, termed the Hydrology Laboratory-Research Distributed Hydrologic Model (HL-RDHM). Because channels are the only source of water exchange between neighboring computational elements, the absence of such exchange has been identified as a weakness in the model. The primary objective of this paper is to modify the model structure to account for subsurface water exchanges without dramatically altering the conceptual framework of the water balance module. The subsurface exchanges are established by partitioning the slow response components released from the lower layer storages into two parts: the first part involves the grid's conceptual channel, while the second is added to the lower layer storages of the downstream pixel. Realizing the deficiency of the water balance module to locate the lower zone layers in sufficient depths, a complementary study is conducted to test the feasibility of further improvement in the modified model by equally shifting downward the lower zone layers of all pixels over the basin. The Baron Fork at Eldon, Oklahoma, is chosen as the test basin. Ten years of grid-based multisensor precipitation data are used to investigate the effects of the modification, plus shifting the lower zone layers on model performance. The results show that the modified-shifted HL-RDHM can markedly improve the streamflow simulations at the interior point, as well as very high peak-flow simulations at the basin's outlet. Copyright 2011 by the American Geophysical Union.