Spectral distributions of auroral optical emissions, peaked at distinctly different heights in the thermosphere, show significant variations, with altitude, in the O2 atmospheric (1, 1)/(0, 1) band ratio. The latter increases with height in auroras peaked between 110 and 150 km and then gradually decreases at higher altitudes. To minimize ambiguities associated with auroral height determination needed for investigating this effect, four independent height-assessment methods are employed. The first one is based on the incoherent scatter radar (ISR) soundings of the auroral ionization profile from which the height, where precipitating particles dissipate most of their energy, can be determined. Concurrent spectroscopic observations of the thermalized rotational distributions of auroral band emissions yield the ambient air temperature, and hence an independent assessment of the height, of the thermospheric region where these emissions peak. Changes in O/O2 and O/N2 ratios with height lead to changes in the ratios of auroral emissions, from these species, peaked at different heights. Finally, changes in collision frequency with height lead to changes in the brightness of the auroral emissions, resulting from radiatively allowed transitions relative to those produced from radiatively forbidden transitions. The four methods yield comparable values for the height of the thermospheric region where emissions, from each auroral event, peak. The observed variations in O2 atmospheric (1, 1)/(0, 1) with auroral height is compared with that expected from O(1D)+O2 excitation source and quenching by O2 and O. The effects of electron impact excitation of O2(b1∑g+,&ngr;') and high rotational levels of the P branch of O2 atmospheric (0, 0) band on O2 atmospheric (1, 1)/(0, 1) ratio are discussed. Quantitative ratios of various auroral emissions, from O, N2, and N2+, peaked at different heights, that can provide an assessment of auroral heights where these emissions peak, are listed. ¿ 1999 American Geophysical Union