This paper computes the F-region winds expected from ion motion and neutral air pressure gradients, and reconciles these with the belief, based on satellite observations, that the upper atmosphere rotates faster than the Earth. The paper points out that the satellite observations refer to wind momentum flux rather than speed, and are weighted to low latitudes and daytime. Here indeed eastward winds are expected; maximumeastward winds are predicted in the early evening, in agreement with satellite findings. But the predicted global wind pattern is not eastward. During the day when ion densities are great, eleetrodynamic drag between the ions and neutral molecules is important. Under these conditions, a strong eastward wind is expected near the equator, where ion velocities are greatest. At night, however, when eleetrodynamic drag is weak, there is no corresponding equatorial westward wind and pressure gradients take control at all latitudes. These cause a strong equatorward wind near midnight, which could have the same effect on a satellite orbit as a weaker eastward wind.