The Mass Spectrometer and Incoherent Scatter (MSIS) series of models has been a standard for specification and prediction of conditions in the Earth's thermosphere. One aspect of the model that has not been well validated is its predictions of the lower thermospheric compositional response to varying periodic and transient solar and geomagnetic forcings. Here we compare the predictions by the NRLMSIS model over the years 2002 to 2006 for column O/N2 with GUVI satellite (daytime) observations at a fixed longitude of the Poker Flat Research Range in central Alaska. NRLMSIS predictions are also compared with nightside auroral O/N2 observations obtained from ground‐based photometric measurements from central Alaska. While there is often good agreement between the GUVI and NRLMSIS O/N2 at high latitudes, there are also times during large geomagnetic storms when there is some disagreement with the largest occurring at midlatitudes. The 9‐day periodicity, first described in 2007 and attributed to high‐speed solar wind streams, is present in the model at all latitudes but is quite weak in the observations below 55N latitude. Away from high‐latitudes, these differences may exist because the NRLMSIS O/N2 dependence on ap forces nonperiodic contributions contained in the NRLMSIS database to give a periodic response when ap is varied periodically. While the nighttime observations show considerable agreement with NRLMSIS they are much more variable than predicted by the model. This variability suggests that there are significant local effects in the auroral zone that are not captured by the model.
