A large equinoctial asymmetry has been observed in thermospheric winds and ion velocities at high latitude sites in northern Scandinavia. Throughout the solar cycle, average nighttime thermospheric meridional winds are larger in spring than autumn despite similar levels of solar insolation. The average ion velocities are also larger in spring than autumn at solar maximum, but at solar minimum this position is reversed. Numerical simulations of the thermosphere and ionosphere have not predicted such asymmetries because they generally assume forcing functions that are symmetric about the solstices. The proposed explanation lies in the annual and diurnal variation in solar wind-magnetosphere coupling caused by changes in the orientation of the geomagnetic pole, and hence the magnetosphere, with respect to the average orientation of the IMF (the Russell-McPherron effect). This causes a 12-hour phase difference between the times of maximum solar wind-magnetosphere coupling at the two equinoxes. In addition, the orientation of the geomagnetic axis with respect to the average IMF is such that ⟨By*Bz⟩>0 for the March equinox and ⟨By*Bz⟩<0 for September. This results in a further source of asymmetry of forcing of the high-latitude ionosphere as the result of electric fields associated with the four sign combinations of By and Bz. Several predictions arise from the explanation given: for example, a high-latitude station measuring thermospheric neutral winds in Alaska, 180¿ in longitude from Kiruna, might be expected to see nighttime thermospheric winds that are larger in the autumn than in the spring. ¿ American Geophysical Union 1996