The goal of the Atmospheric Response in Aurora (ARIA) experiment carried out at Poker Flat, Alaska, on March 3, 1992, was to determine the response of the neutral atmosphere to the long-lived, large-scale forcing that is characteristic of the diffuse aurora in the postmidnight sector. A combination of chemical release rocket wind measurements, instrumented rocket composition measurements, and ground-based optical measurements were used to characterize the response of the neutral atmosphere. The rocket measurements were made at the end of a 90-min period of strong Joule heating. We focus on the neutral wind measurements made with the rocket. The forcing was determined by running the assimilated mapping of ionospheric electrodynamics (AMIE) analysis procedure developed at the National Center for Atmospheric Research. The winds expected at the latitude and longitude of the experiment were calculated using the spectral thermospheric general circulation model developed at the Danish Meteorological Institute. Comparisons of the observations and the model suggest that the neutral winds responded strongly in two height ranges. An eastward wind perturbation of ~100 m s-1 developed between 140 and 200 km altitude with a peak near 160 km. A southwestward wind with peak magnitude of ~150 m s-1 developed near 115 km altitude. The large amplitude winds at the lower altitude are particularly surprising. They appear to be associated with the upward propagating semidiurnal tide. However, the amplitude is much larger than predicted by any of the tidal models, and the shear found just below the peak in the winds was nominally unstable with a Richardson number of ~0.08. ¿ American Geophysical Union 1995