We investigate the impact of climate change on future air quality in the United States with a coupled global/regional scale modeling system. Regional climate model scenarios developed by dynamically downscaling outputs from the GISS GCM are used by CMAQ to simulate present air pollution climatology, and modeled surface ozone mixing ratios are compared with recent observations. Though the model accurately simulates ozone in the northeast U.S. and in central California, a positive bias of 10-15 ppb exists throughout most of the central and southeast U.S. The model is also applied to a simulated 2050 climate based on the IPCC AlB greenhouse gas scenario. Two future simulations are conducted, one with anthropogenic emissions held at 2001 levels, and one with anthropogenic emissions reduced in accordance with the AlB scenario. Without ozone precursor emissions changes, increases from 2-5 ppb in summer mean 8-h ozone mixing ratios are projected in Texas and parts of the eastern U.S., while high ozone episodes become more frequent. Increases of 2-8 ppb during the autumn are predicted over a large area in the central and southwest U.S., suggesting a lengthening of the ozone season. These increases within the regional modeling domain are predicted despite large decreases in the future global background ozone mixing ratio. Substantial decreases exceeding 15 ppb during the summer are predicted for the future reduced emissions case. A sensitivity test conducted with 30% higher methane mixing ratio yields widespread ozone increases of 0.5-2 ppb, an effect larger than that of climate-induced increases in isoprene emissions, demonstrating the need to consider changes in methane levels alongside climate change when simulating future air quality Copyright 2008 by the American Geophysical Union.