Quantitative measurements of historic improvements in fuel efficiency technology help to illuminate the feasibility of future fuel economy standards. Past investigations have produced widely varying estimates of this rate of improvement, though all seem to indicate that fuel consumption reductions implied by the 2025 U.S. CAFE standards cannot be met solely through technological improvements at historic rates. In this paper, we use the characteristics of U.S. cars between 1975 and 2009 to estimate that holding all else equal, a 1% increase in weight increases a car's fuel consumption by 0.69%, and a 1% reduction in 097 km/h acceleration time increases fuel consumption by 0.44%. These tradeoff parameters are combined with the results of related work by the authors and others, yielding a more comprehensive measure of technological improvements than has been previously reported. When accounting for all of these sources of improvement, we conclude that the per-mile (or per-kilometer) fuel consumption of new cars in the U.S. could have been reduced by 5% per year from 1975 to 1990, if acceleration, features, and functionality had remained at their 1975 levels. Approximately 80% of this potential was realized as actual reductions in fuel consumption. Between 1990 and 2009, in contrast, technological improvement averaged just 2.1% per year, only 34% of which was realized as actual fuel consumption reductions. To meet the 2025 CAFE standards for cars without sacrificing capabilities that consumers have come to expect, technology must improve quickly enough to reduce fuel consumption by 4.3% per year for 14 years considerably faster than has occurred since 1990, but consistent with the pace of improvements observed between 1975 and 1990. (C) 2015 Elsevier Ltd. All rights reserved.