We have used two different sets of assimilated input wind fields (United Kingdom Meteorological Office (UKMO), NASA) to drive the transport in a three-dimensional global chemical transport model. In a first experiment the distribution of carbon 14 after nuclear bomb testing in the early 1960s was simulated. The second, more comprehensive experiment consists of a comparison of simulated nitrous oxide (N2O) distributions for both input wind fields against CLAES satellite measurements. Both experiments show that the NASA winds are producing too much upward transport in the equatorial area. By analyzing the Eulerian-mean conservation equation for zonal mean mixing ratios, we were able to show that these model differences are initially caused by differences in the mean residual circulations in the two different wind field sets but that because the latitudinal gradients remain steeper in the NASA results, differences in isentropic mixing rates are also evident. The NASA residual mean circulation in the equatorial lower stratosphere in the model is higher than the UKMO residual mean circulation by about a factor of 2. However, there is also evidence that isentropic mixing rates are too large in the subtropics in both models, and there is evidence from the N2O comparisons of overly large amplitudes of wavenumbers 4 to 7 in the tropics and subtropics in the models. Filtering the incoming wind fields to eliminate wavenumbers larger than 10 resulted in only minor improvements in the model comparisons against CLAES observations. Because of the significant discrepancies between the model simulations, using either one of the wind field sets, and the observations, we recommend care in using these assimilated winds for long-term model studies in general. Both wind field sets, however, produced good simulations of strong intrusions of tropical air into midlatitudes of the upper stratosphere in September 1992. ¿ 1998 American Geophysical Union