The Stokes drift argument put forth by Coy et al. (1986) to explain the monthly mean downward vertical velocities observed in the summer polar mesosphere by radars is critically re-examined. The size of the effect is quite sensitive to the choice of gravity wave wavenumber and phase speed. We reproduce the Coy et al. (1986) result for a monochromatic gravity wave and then generalize to a Garrett-Munk type spectrum. This allows us to easily incorporate experimentally determined wave field parameters in order to predict a Stokes drift magnitude. The Stokes drift we calculate for mesospheric spectra reported in the literature is less than 4 cm/s. This is nearly a factor of 10 smaller than the mean June and July downward vertical velocities we have verified from 4-year averaging of the Poker Flat, Alaska, radar data base. In this extended analysis we also find that the upward winter mean vertical velocity is smaller than previously reported but still in apparent conflict with the mesospheric winter circulation theory. We suggest, as an alternative to the Stokes drift idea, that the observed summer velocity might be the terminal velocity of charged aerosols. Such an explanation would have the advantage of relating the mean velocity observations to the large VHF radar cross section exhibited by the polar summer mesosphere. This possible connection is discussed further in a companion paper. ¿ American Geophysical Union 1992