A full-wave dynamical model and chemistry models that simulate ground-based observations of gravity wave-driven O2 atmospheric and O I 5577 airglow fluctuations in the mesopause region are used to demonstrate that for many observable gravity waves modeling is required to infer temperature perturbation amplitudes from airglow observations. We demonstrate that the amplitude of the altitude-integrated volume emission rate weighted temperature perturbation differs by at least about 30% from the amplitude of the temperature perturbation of the major gas in the vicinity of the peak of the airglow volume emission rate for gravity waves with horizontal phase speeds less than about 150 m s-1 and vertical wavelengths less than about 50 km and that the amplitude of the altitude-integrated volume emission rate weighted temperature perturbation differs considerably from the amplitude of the temperature perturbation averaged over the vertical extent of the emission layer for waves with horizontal phase speeds less than about 65 m s-1 and vertical wavelengths less than about 20 km. For waves with phase speeds less than about 100 m s-1 and vertical wavelengths less than about 30 km the amplitude of the altitude-integrated volume emission rate weighted temperature perturbation differs by at least about 30% from the altitude-integrated mean volume emission rate weighted temperature perturbation, demonstrating that the nonthermal fluctuation contribution to the former (involving volume emission rate perturbations) needs to be included in such modeling. We conjecture that the observed brightness perturbation is a simpler and better quantity to simulate using detailed modeling than the observed airglow temperature perturbation for the determination of wave amplitude in cases where nonthermal effects or cancellation effects (for short vertical wavelengths) are strong. ¿ 1999 American Geophysical Union