Driver behavior has an immense impact on vehicle fuel economy and emissions, yet it has historically been treated as random error in models of fuel economy and neglected in energy and environmental policy-making regarding fuel efficiency. Recently, concern about fossil fuel depletion and climate change, as well as the critical role of driver behavior in achieving the fuel economy benefits of new hybrid and electric vehicles, has created interest in eco-driving. Eco-driving refers to suites of behavior a driver can engage in to improve fuel economy. The most common strategy used to promote eco-driving is feedback that conveys information about fuel efficiency to the driver. Feedback is typically visual and provided on-board the vehicle via digital screens (dash or instrument cluster displays, after-market devices, or web apps on personal smartphones or tablets). No policies exist requiring manufacturers to provide eco-driving feedback, yet feedback systems of increasing variety are appearing in vehicles, likely due to advances in telematics and decreasing costs of new technologies. The rapidly increasing prevalence and complexity of in-vehicle information systems, along with concern for driver distraction, suggest standardization of eco-driving feedback may be warranted in the near future. Thus, there is a need to understand what types of eco-driving feedback are effective. Although the literature generally suggests that feedback can be effective in supporting eco-driving, results are widely variable. This inconsistency is likely due to variation among studies, e.g., in terms of feedback design, sample, setting, and length of intervention. For example, many studies use vehicle simulators, while others outfit participants’ personal or work vehicles with feedback devices. Some involve professional drivers of fleet vehicles and others civilian drivers of private vehicles. Feedback design ranges widely, for example, from haptic accelerator pedals that create resistance when the driver attempts to over-speed, to complex visual displays that gamify fuel economy, rewarding the driver with points or growing trees. This white paper presents a statistical meta-analysis of eco-driving feedback studies in order to determine a pooled estimate of the impact on fuel economy and explore how characteristics of feedback interventions influence their impact. This review is for policy-makers and fleet operators who have a stake in reducing vehicle fuel consumption and emissions. It provides the most accurate estimate to-date of the average impact of in-vehicle feedback on fuel economy and summarizes the current state of knowledge regarding characteristics of eco-driving feedback interventions that determine effectiveness. A literature search was conducted for studies with experimental designs that measured the impact of onboard feedback on objective eco-driving outcomes. Studies were coded on variables related to feedback design, intervention, and study characteristics. The meta-analysis was conducted using the metasem package for R statistical software, using a random effects, three-level model. Outcome measurements were translated into a common effect size: relative iii (i.e., percent) improvement in fuel economy. After excluding studies that were not amenable to the analysis, the final sample size included 23 effects in 17 studies. The main effect of onboard feedback on fuel economy across all 17 studies and 23 effect sizes was 6.6% improvement (with 95% confidence that the true population effect would fall between 4.9% and 8.3%). The average fuel economy without feedback in these studies (i.e., in baseline phases or control groups) was about 25 MPG. A 6.6% improvement from this baseline would be equivalent to a 1.7 MPG improvement. The rate at which Corporate Average Fuel Economy (CAFE) standards are progressing may slow considerably under the Trump administration. Eco-driving feedback is a strategy that enhances consumers’ own control over their fuel economy, which may align better with conservative ideology. This study also tested fourteen hypotheses, grounded in behavioral theory and past empirical research, about the characteristics and contexts that make eco-driving feedback more effective. Only one hypothesized relationship emerged as statistically significant at the alpha = .10 level; this was the negative relationship between length of intervention (i.e., number of days drivers were exposed to feedback) and effect size. On average, the effect of feedback decreased as length of intervention increased, suggesting eco-driving feedback programs and technologies should not count on persistent savings and should assess program costs accordingly. Given that eco-driving feedback outcomes are generally better in the short-term, it is crucial to understand how feedback design can maximize and prolong effects. Likely due to small sample sizes, feedback design variables did not emerge as statistically significant moderators of effectiveness. However, trends in these variables aligned with study hypotheses, suggesting feedback should: (a) be provided in multiple modalities (e.g., visual and haptic or auditory rather than visual only); (b) include both fine- and course-grained information; (c) provide feedback standards against which to compare performance; (d) integrate gameful design elements (e.g., points, levels, badges); and (e) be combined with other interventions, such as education and rewards contingent on performance. More experiments that compare the impact of different feedback designs are needed in order to identify the most promising designs, which can then be promoted to manufacturers and inform potential future standardization of fuel economy and related displays.