An experimental procedure for detecting small changes in automotive fuel economy via on-road testing is described. The approach uses standard statistical methods for experimental design. The procedure is demonstrated by measuring the change in fuel economy caused by using air conditioning and by opening side windows. Analysis of results for two test vehicles shows that the air conditioning effect is significantly greater than that of open windows for both vehicles. This result is consistent with estimates based on the best available information, and is supported by recent results from other similar studies.