There is currently much interest in fluctuations of airglow emissions caused by atmospheric gravity waves. The fluctuations of brightness tend to be found in phase (or occasionally in antiphase) with the fluctuations of measured temperature, whereas current theory tends to anticipate substantial phase differences. We suggest here that the discrepancy results from failure of the common theoretical assumption that the relevant gravity waves are dominated by a single upgoing component: that, instead, there is an accompanying downgoing component of comparable magnitude, produced by reflection. In the case of total reflection, simple, steady state chemistry and vertical viewing, the phase difference is necessarily zero (or 180¿). ¿ American Geophysical Union 1994