We have compared a diurnal photochemical model of ozone with nighttime data from the limb infrared monitor of the stratosphere (LIMS) and ground-based microwave observations. Consistent with previous studies, the model underpredicts the observations by about 10--30%. This agreement is strong confirmation that the model ozone deficit is not simply an artifact of observational error since it is unlikely to occur for two completely different ozone data sets. We have also examined the seasonal, altitudinal, and diurnal morphology of the ozone deficit. Both comparisons show a deficit that peaks in the upper stratosphere (2--3 mbar) and goes through a minimum in the lower mesosphere from 1.0 to 0.4 mbar. At lower pressures (<0.2 mbar) the deficit appears to increase again. The seasonal variation of the deficit is less consistent. The deficit with respect to the LIMS data is least in winter while with respect to the microwave data, the deficit shows little seasonal variation. Finally, the night-to-day ratio in our model is in generally good agreement with that seen in the microwave experiment. Increasing the rate coefficient for the reaction O+O2+M→O3+M improves the fit, while a very large (50%) decrease in HOx catalytic cycle is not consistent with our observations. Increasing the atomic oxygen recombination rate also improves the overall agreement with both data sets; however, a residual discrepancy still remains. There appears to be no single chemical parameter which, when modified, can simultaneously resolve both the stratospheric and mesospheric ozone deficits. ¿ American Geophysical Union 1995