The atmosphere is highly stratified, even to high levels, and hence important transport questions arise in which vertical transport alone constitutes a useful approximation. Thus, the thermosphere was first described in terms of molecular conduction transferring heat from higher levels to lower, which is required because of the lack of adequate energy emission mechanisms at the higher levels. Later, the importance of eddy conduction was recognized as a means of extending the treatment well down into the mesosphere, where finally radiative processes are powerful enough to dispose of the excess heat deposited at the higher levels. Transport problems also arise with regard to atmospheric constituents, and several studies have been made of this problem. These considerations have provided a conceptually complete description of the upper atmosphere in the approximation that horizontal transport can be neglected. Two different approaches have been undertaken to consider the horizontal transport by large-scale circulation. First and most direct, the horizontal wind field in the middle and upper thermosphere was calculated based on the pressure distributions derived from satellite orbital decay observations. Then continuity conditions were used to calculate the vertical motions. The second approach was to evaluate the energy deficits or surpluses as a function of altitude and latitude and to assume the presence of vertical motions sufficient to balance these deficits or surpluses by compressional heating or cooling. Then continuity conditions were used to derive horizontal winds. The results of the two approaches are compatible, and are largely complementary. Downward velocities are near 1 m s-1 at 300 km over the diurnal minimum and the winter polar region, and 1 cm s-1 at 100 km over the winter polar region.