Very good agreement is shown for diurnal and semidiurnal temperature variations calculated from lidar measurements in southern France and from data of the microwave limb sounder of the Upper Atmosphere Research Satellite (UARS). Tides induce temperature deviations observed in southern France to be as large as ¿3 K, with a maximum at the stratopause. The amplitudes and phases of the semidiurnal variation change significantly with season and location. Seasonal changes up to 2 K have been clearly identified for the diurnal component. An analytic model of the diurnal component, based on sinusoidal functions, fits the data well, but is less successful for the semidiurnal component. Substantial agreement is also reported for the diurnal component between the results of our analytical model and the published results of a two-dimensional global-scale wave model. In contrast, the semidiurnal component is in total disagreement with numerical simulations that report very small amplitudes, as compared with the observations reported here. The confidence in detecting bias in data comparisons is improved if data used are limited to periods from April to September and if time-of-day adjustments are applied. Comparison between lidar and nearly coincident UARS temperature measurements have revealed, systematically, for the 4 experiments aboard UARS, a significant residual mean difference of up to 3 K around 35--43 km. A comparison using simultaneous measurements suggests that the bias is associated with the variability of migrating tides and/or the presence of nonmigrating tides rather than instrumental characteristics. ¿ American Geophysical Union 1996