We present first results from a comparison of an extensive set of ionospheric storm measurements with a comprehensive model that includes interhemispheric coupling, photoelectrons, vibrationally excited N2, and the MSIS-86 neutral atmosphere model. While the modeled electron density and temperature agree well with the measured density and temperature for the quiet day, the model does not reproduce the observed factor of 1.7 decrease in the electron density on the storm day. The model does not support a recent suggestion that vibrationally excited N2 might account for the negative phase of ionospheric storms. The most likely cause of the model failure is an inadequate decrease in the ratio of atomic oxygen to molecular neutral densities in the MSIS neutral atmospheric model for this particular storm. A factor of 3 to 5 decrease in the atomic oxygen density to molecular density ratio at 300 km is needed to explain the observed decrease in electron density. This result is important because (1) it indicates that the new MSIS model may not be reliable during particular magnetic storms and (2) it affects the accuracy of neutral winds calculated from radar measurements of the ion drift and also winds derived from the height of the F2 layer peak density. ¿ American Geophysical Union 1989