This paper is primarily concerned with the causes of the large density and temperature enhancements that are often observed during magnetically quiet periods on winter nights at midlatitudes in the North American sector. Measurements from a network of Digisondes and an incoherent scatter radar are compared with the field line interhemispheric plasma (FLIP) model for January 6--12, 1997, in order to examine the temporal evolution and geographical extent of the enhancements in eastern North America. Postsunset measurements at Millstone Hill show high electron temperatures accompanied by rapid density decay until midnight followed by a rapid temperature decay accompanied by a pronounced density enhancement in the early morning hours. The FLIP model reproduces the nighttime density enhancement well, provided the model is constrained to follow the topside electron temperature and also that the overlying plasmaspheric flux tube is full. The dramatic reduction in plasmaspheric heat flux near midnight results in a sharp decrease in ionospheric temperature, inducing a large downward flow of plasmaspheric ions which creates the nighttime enhancement in ionospheric density. We find that the nighttime plasmaspheric heat flux variation drives the nighttime ionospheric density variation, which is opposite to conclusions in previously published work. Although the plasmaspheric heat flux variation can explain the ionospheric density variation, the reasons for this heat flux variation are not understood. Convection of plasma from higher magnetic latitudes is now included in the FLIP model but is not needed to produce the observed nighttime density maximum. We have found that the fraction of light ions in the topside ionosphere at 500 km altitude in the model is very close to that obtained from chemical equilibrium and agrees well with the measured fraction. The model generally reproduces the daytime electron density very well at all stations except Bermuda, where the difference is as much as 50%. ¿ 2000 American Geophysical Union