The effects of an artificial upper boundary on the climate of a general circulation model are examined using two versions of the model. The complete model has 26 levels extending from the surface to 0.1 mb (∼65 km) and the second version approximates the usual tropospheric general circulation model, having 15 levels extending to 10 mb (∼30 km). The two model versions are identical over the common part of the domain. Vertically propagating Kelvin waves are found in the tropics of both versions of the model. In the version extending only to 10 mb, the Kelvin waves are reflected by the top boundary, and resemble standing oscillations in the vertical. while retaining (approximately) their original zonal phase speed. Some of these reflected waves develop much larger lower stratospheric amplitudes than when they are able to propagate freely into the mesosphere. Placing the topmost model level at 10 mb is found to radically alter the winter stratospheric circulation and to produce significant changes in... Abstract The effects of an artificial upper boundary on the climate of a general circulation model are examined using two versions of the model. The complete model has 26 levels extending from the surface to 0.1 mb (∼65 km) and the second version approximates the usual tropospheric general circulation model, having 15 levels extending to 10 mb (∼30 km). The two model versions are identical over the common part of the domain. Vertically propagating Kelvin waves are found in the tropics of both versions of the model. In the version extending only to 10 mb, the Kelvin waves are reflected by the top boundary, and resemble standing oscillations in the vertical. while retaining (approximately) their original zonal phase speed. Some of these reflected waves develop much larger lower stratospheric amplitudes than when they are able to propagate freely into the mesosphere. Placing the topmost model level at 10 mb is found to radically alter the winter stratospheric circulation and to produce significant changes in...