This paper examines the ability of empirical and physical models to reproduce the peak electron density of the midlatitude ionospheric F2 region (NmF2) from 1976 to 1980. The data from all midlatitude stations show a tendency toward a semiannual variation in noon NmF2 with peaks at the equinoxes for all levels of solar activity. The Southern Hemisphere semiannual variation is more pronounced than in the Northern Hemisphere primarily because the winter density is relatively low in the Southern Hemisphere. At most locations the equinox density peaks are approximately equal. However, the September peak is much weaker than the March peak at most Australian stations. This leads to a distinct longitudinal variation between the Australian and South American sectors. On the other hand, there is remarkably little longitudinal variation in the Northern Hemisphere. We present calculations from the field line interhemispheric plasma (FLIP) model from 1976 to 1980 at six representative midlatitude stations around the globe. The FLIP model reproduces the average seasonal and solar cyclical behavior of the measured NmF2 remarkably well most of the time. The greatest differences of 50% occur at the March equinox in the South American region and at the September equinox in the Australian region during September solstice solar maximum. The international reference ionosphere (IRI) model reproduces the average NmF2 even better than the FLIP model but, unlike the FLIP model, it has little day-to-day variation. A factor of 2 increase in the solar EUV ion production rate and in the atomic to molecular density ratio at the F2 region peak height (hmF2) produces a factor of 4 increase in NmF2 over a solar cycle. Most of this increase takes place before the average solar activity index (F10.7) reaches 175. At solar maximum in 1979 and 1980, there is little relationship between daily F10.7 and NmF2. Changes in the atomic to molecular density ratio at hmF2 are primarily responsible for the semiannual variation in the FLIP model NmF2. The inclusionof vibrationally excited N2 in the FLIP model improves the relative seasonal and solar cycle NmF2 variations in the FLIP model, but it causes the overall NmF2 to be too low at most stations. ¿ 2001 American Geophysical Union