A numerical full-wave model is used to study the response of the O2 atmospheric airglow to ducted gravity waves in the mesopause region. For an isothermal, quasi-adiabatic, and motionless background atmosphere the calculated phase differences between airglow brightness fluctuations and fluctuations of temperatures derived from the airglow, as given by Krassovsky's ratio, are in good agreement with the predictions of published theory. Significant departures from the predictions of the basic theory are obtained when we consider ducting in the presence of the eddy and molecular diffusion of heat and momentum in a nonisothermal background atmosphere. Wind shears also affect the phase difference between airglow brightness fluctuations and temperatures derived therefrom. Nonisothermal affects and the effects of diffusion and winds are largest for the slower waves we consider. Only the fastest of the ducted waves considered conform to the basic theory, while the airglow signatures associated with slower, more weakly ducted waves may be easily misinterpreted as being due to propagating waves. We conclude that for the short horizontal wavelength waves observed in the airglow, the phase of Krassovsky's ratio may be useful to identify wave ducting only for the shortest period, fastest waves. Therefore identification of ducted waves using Krassovsky's ratio will be difficult even if the required high temporal resolution measurements become available. ¿ 2001 American Geophysical Union