A multiple linear regression statistical method is applied to model data taken from the Coupled Model Intercomparison Project, phase 5 (CMIP-5) to estimate the 11-year solar cycle responses of stratospheric ozone, temperature, and zonal wind during the 1979-2005 period. The analysis is limited to the six CMIP-5 models which resolve the stratosphere (high-top models) and which include interactive ozone chemistry. All simulations assumed a conservative 11-year solar spectral irradiance (SSI) variation based on the Naval Research Laboratory model. These model responses are then compared to corresponding observational estimates derived from two independent satellite ozone profile datasets and from ERA-Interim reanalysis meteorological data. The models exhibit a range of 11-year responses with three models (CESM1-WACCM, MIROC-ESM-CHEM and MRI-ESM1) yielding substantial solar-induced ozone changes in the upper stratosphere which compare favourably with available observations. The remaining three models do not, apparently because of differences in the details of their radiation and photolysis rate codes. During winter in both hemispheres, the three models with stronger upper-stratospheric ozone responses produce relatively strong latitudinal gradients of ozone and temperature in the upper stratosphere which are associated with accelerations of the polar night jet under solar maximum conditions. This behaviour is similar to that found in the satellite ozone and ERA-Interim data, except that the latitudinal gradients tend to occur at somewhat higher latitudes in the models. The sharp ozone gradients are dynamical in origin and assist in radiatively enhancing the temperature gradients, leading to a stronger zonal wind response. These results suggest that simulation of a realistic solar-induced variation of upper-stratospheric ozone, temperature and zonal wind in winter is possible for at least some coupled climate models even if a conservative SSI variation is adopted.