Primitive equation and quasi-geostrophic eddy resolving, open ocean models are used for hindcast studies in the Gulf Stream meander and ring formation region. A feature model approach is used to initialize the models, based on one month of observations during November to December 1984. Flat bottom and topographic calculations are carried out using an initial Gulf Stream velocity profile based on the Pegasus dataset. All of the major events observed in the upper thermocline are reproduced by both numerical models. The addition of bottom topography is shown to significantly alter the character of the deep velocity fields. Large, basin scale circulations found near the bottom in both flat bottom calculations were replaced by energetic jets and eddies associated with the dominant spatial scales of the bottom topography. Use of the quasi-geostrophic model to dynamically adjust the initial conditions for the primitive equation model is shown to reduce the growth of large scale meanders on time scales o... Abstract Primitive equation and quasi-geostrophic eddy resolving, open ocean models are used for hindcast studies in the Gulf Stream meander and ring formation region. A feature model approach is used to initialize the models, based on one month of observations during November to December 1984. Flat bottom and topographic calculations are carried out using an initial Gulf Stream velocity profile based on the Pegasus dataset. All of the major events observed in the upper thermocline are reproduced by both numerical models. The addition of bottom topography is shown to significantly alter the character of the deep velocity fields. Large, basin scale circulations found near the bottom in both flat bottom calculations were replaced by energetic jets and eddies associated with the dominant spatial scales of the bottom topography. Use of the quasi-geostrophic model to dynamically adjust the initial conditions for the primitive equation model is shown to reduce the growth of large scale meanders on time scales o...