Model simulations of temperature and density trends in the upper thermosphere are generally consistent with satellite drag data, but some discrepancies remain. The most important of these is that satellite drag analyses under solar minimum conditions have measured density change of about À5% per decade near 400 km altitude, while model simulations of upper atmosphere cooling due to anthropogenic increases in carbon dioxide and other trace gases have predicted about half that rate. For solar moderate and maximum conditions, agreement is better. The rate of change is less during higher solar activity, because higher levels of nitric oxide cooling compete with the anthropogenic cooling. However, some past modeling studies used global mean models, and others attempted to scale to decadal rates from scenarios where carbon dioxide was doubled. Both of these approaches have shortcomings. Therefore, we have performed new, fully 3-D simulations, using the National Center for Atmospheric Research thermosphere-ionosphere-mesosphere electrodynamics general circulation model, to better quantify secular change rates at various levels of solar activity. These simulations use a 12 year baseline (approximately one solar cycle) in order to more directly compare with measured rates. Our new findings are in better agreement with observations for solar minimum conditions, approximately À5% per decade at 400 km, and are also still in reasonable agreement at solar maximum, approximately À2% per decade. This confluence of observation and simulation strengthens the case that some of the best evidence of the impact of anthropogenic global change on the upper atmosphere is the continued systematic decrease of thermospheric density.