The U.S. Environmental Protection Agency (EPA) presently uses a 40 ppbv background O3 level as a baseline in its O3 risk assessments. This background is defined as those concentrations that would exist in the absence of North American emissions. Lefohn et al. <2001> have argued that frequent occurrences of O3 concentrations above 50--60 ppbv at remote northern U.S. sites in spring are of stratospheric origin, challenging the EPA background estimate and implying that the current O3 standard (84 ppbv, 8-hour average) may be unattainable. We show that a 3-D global model of tropospheric chemistry reproduces much of the observed variability in U.S. surface O3 concentrations, including the springtime high-O3 events, with only a minor stratospheric contribution (always 2 km). It declines from spring to summer and further decreases during O3 pollution episodes. The 40 ppbv background assumed by EPA thus actually underestimates the risk associated with O3 during polluted conditions. A better definition would represent background as a function of season, altitude, and total surface O3 concentration. Natural O3 levels are typically 10--25 ppbv and never exceed 40 ppbv. International controls to reduce the hemispheric pollution background would facilitate compliance with an AOT40-type standard (cumulative exposure to O3 above 40 ppbv) in the United States.