Measurements of the Doppler width of the 6300 ¿ airglow emission line have been extensively used to determine the thermospheric temperature. This technique is based on the assumption that the bulk of the emitting O(1D) atoms are thermalized in the region of the airglow source (200--300 km). A Monte Carlo stochastic model is used to calculate the energy distribution function of O(1D) atoms in the daytime and nighttime thermosphere. Hot O(1D) atoms are produced by exothermic processes and their thermalization is controlled by the competition between radiation, collisional quenching, and relaxation. It is found that the O(1D) temperature departs from the background gas temperature not only in the upper thermosphere but also in the region of the bulk 6300 ¿ emission. At 300 km for low solar activity conditions, the model predicts an excess O(1D) temperature of ~180 K during daytime and ~950 K at night. The temperature departure persists at lower altitudes as a result of the major contribution of the O2+ dissociative recombination source of hot 1D atoms. Experimental evidence based on the Fabry-Perot interferometer measurements on board the Dynamics Explorer satellite confirms the existence of an O(1D) temperature excess over the mass spectrometer/incoherent scatter (MSIS) value. It is concluded that temperatures deduced from the 6300 ¿ airglow line width may significantly exceed the ambient gas temperature in a way depending on solar activity, local time, and observation geometry. ¿ 1999 American Geophysical Union