The effect of gasoline olefin composition and content on urban ozone was estimated using the Urban Airshed Model (UAM), emission measurements for a base fuel, and projected emissions for two hypothetical fuels with reduced olefin content. The projected emissions for the hypothetical fuels were developed using regressions developed from Auto/Oil Air Quality Improvement Research Program (AQIRP) Phase I testing, a vapor headspace model and other information. Ozone modeling was conducted for Los Angeles in year 2010 and Dallas-Fort Worth and New York in year 2005. When all olefins were removed from the base fuel, the light-duty vehicle contribution to peak hourly ozone was reduced by 8 to 12%. This corresponds to a projected reduction of 0.6 to 0.8% in total peak ozone from all sources. Removing only light (C5) olefins provided 67 to 78% of the peak ozone benefit from removal of all olefins. Reducing fuel C5 olefin content was estimated to be 3.5 to 6 times more effective in reducing peak ozone than reducing heavier olefins (C6+) by an equal volume percent. The effect of olefins on predicted ozone was due primarily to a reduction in the reactivity of evaporative, running loss and refueling/storage emissions. There is uncertainty in the predicted effects; in particular, the effect of olefin reduction expressed as a percentage of total peak ozone from all sources may be underestimated.