A simple, free-surface, barotropic model and a nine-level, baroclinic model are numerically time integrated on both latitude-longitude grids and on Kurihara-type grids to compare the results obtained from the two grid systems. The prognostic variables are Fourier space-filtered in the longitudinal direction on the latitude-longitude grids to permit the use of the same time-step length on both grids. With respect to geopotential height and zonal wind distributions and to the phase speed of wave propagation, the results from the barotropic model, time-integrated on a sector latitude-longitude grid, agree better with a high-resolution control run than those computed on a modified Kurihara grid, particularly at high latitudes. The barotropic model is also time-integrated on a hemispheric, latitude-longitude grid, and the results compare well with a high-resolution control. The latter comparison is performed on initial data having strong cross-polar flow. The mean sea-level pressure distribution obtai... Abstract A simple, free-surface, barotropic model and a nine-level, baroclinic model are numerically time integrated on both latitude-longitude grids and on Kurihara-type grids to compare the results obtained from the two grid systems. The prognostic variables are Fourier space-filtered in the longitudinal direction on the latitude-longitude grids to permit the use of the same time-step length on both grids. With respect to geopotential height and zonal wind distributions and to the phase speed of wave propagation, the results from the barotropic model, time-integrated on a sector latitude-longitude grid, agree better with a high-resolution control run than those computed on a modified Kurihara grid, particularly at high latitudes. The barotropic model is also time-integrated on a hemispheric, latitude-longitude grid, and the results compare well with a high-resolution control. The latter comparison is performed on initial data having strong cross-polar flow. The mean sea-level pressure distribution obtai...