The National Center for Atmospheric Research (NCAR) thermospheric general circulation model (TGCM) for the Earth's thermosphere has been modified to examine the three-dimensional (3D) structure and circulation of the upper mesosphere and thermosphere of Venus (VTGCM). This model used the parameterizations from our earlier two-dimensional (2D) Venus model, including eddy diffusion and wave-drag parameterizations, and the 15-μm cooling scheme from Bougher et al. (S W. Bougher, R. E. Dickinson, E. C. Ridley, R. G. Roble, A. F. Nagy, and T.E. Cravens 1986, Icarus 68, 284–312). Processes unique to the Earth's thermosphere (ion drag, magnetospheric convection, etc.) are removed, but the TGCM computational framework is retained. A symmetric version of the VTGCM is used first to simulate the mean subsolar-to-antisolar variation of observed composition and temperatures, as determined from Pioneer Venus data and subsequent emperical models. The VTGCM equatorial fields are shown to be largely consistent with previous symmetric 2D model fields of Bougher et. al. (S. W. Bougher, R. E. Dickinson, E. C. Ridley, R. G. Roble, A. F. Nagy, and T. E. Cravens 1986, Icarus 68, 284–312). The VTGCM is then used to examine the 3D character of the Venus asymmetric circulation and structure. A prescribed Venus retrograde zonal wind profile has been superimposed on the mean subsolar-to-antisolar circulation, and results show that the major (heavy) species and temperatures are not greatly affected by this superrotation. A shift in the exospheric temperature minimum to LT = 2 AM occurs, which is consistent with the convergence of the horizontal winds after midnight. Many of the observed features of the Venus thermosphere can be reproduced by the 3D VTGCM. Calculated terminator winds reach 230 m sec −, with exospheric temperatures ranging from 309 (day) to 136°K (night). Model atomic oxygen concentrations show little diurnal variation along constant pressure surfaces. Prescribed wave drag is primarily responsible for this weak global circulation, which is consistent with the observed day-night contrast in calculated densities and temperatures. Furthermore, the incorporation of a retrograde zonal momentum source helps to simulate observed asymmetries in Venus thermospheric fields. The present VTGCM thus serves as a useful benchmark upon which to incorporate additional minor constituents and test new self-consistent parameterizations for wave drag and superrotation.