Three-dimensional simulations of the stratospheric constituents CH4, N2O, O3, SF6, and CO2 over an annual cycle have been performed using a semi-Lagrangian chemical transport model <Rasch and Williamson, 1990; Rasch et al., 1994> driven by archived wind data from the Middle Atmosphere version of the National Center for Atmospheric Research Community Climate Model version 2 (MACCM2) general circulation model. The constituents undergo chemical production and loss at rates calculated by two-dimensional photochemical models. We compare these off-line simulations of CH4 and N2O with on-line simulations in which the trace constituent distributions are computed interactively within the MACCM2 general circulation model and find good agreement even when daily averaged wind data and no subgrid scale parameterized mixing are used in the off-line simulations. We also compare the model simulations to satellite, aircraft, and balloon measurements. In most regions and seasons, the zonally averaged model CH4, N2O, and O3 fields agree well with observations. Notable discrepancies are (1) a lack of a double peak structure in the zonally averaged mixing ratios of model CH4 and N2O at equinox, (2) an overall underestimate of CH4 and N2O in the upper stratosphere, and (3) an underestimate of the height of the mixing ratio peak in O3, particularly at high latitudes. We find good agreement between modeled CO2 and SF6 and recent aircraft observations in the lower stratosphere, and balloon measurements in the lower and middle stratosphere. From the SF6 distribution we determine the mean age of air in the model stratosphere, with values as old as 10 years in the wintertime polar upper stratosphere. In addition, we simulate the annual cycle of CO2, a stringent test of model transport, which supplements the mean age. We obtain good agreement with aircraft measurements in phase and magnitude at the tropical tropopause, and the vertical profiles of CO2 are similar to those observed. However, the amplitude of the cycle attenuates too rapidly with height in the model stratosphere, suggesting the influence of midlatitude air and/or the vertical diffusion are too large in the model tropics.¿ 1997 American Geophysical Union