The meteorological predictions of fully coupled air-quality models running in “feedback” versus “no-feedback” simulations were compared against each other and observations as part of Phase 2 of the Air Quality Model Evaluation International Initiative. In the “no-feedback” mode, the aerosol direct and indirect effects were disabled, with the models reverting to either climatologies of aerosol properties, or a no-aerosol weather simulation. In the “feedback” mode, the model-generated aerosols were allowed to modify the radiative transfer and/or cloud formation parameterizations of the respective models. Annual simulations with and without feedbacks were conducted on domains over North America for the years 2006 and 2010, and over Europe for the year 2010.