This paper analyzes the impact of changes in average fuel efficiency on traffic-related fatalities while controlling for other confounding effects. These other effects include population, per capita income, per capita alcohol consumption, existence of safety-belt laws (and safety-belt usage), and age cohorts in the population. State-level time-series data over 24 years is used with a fixed effect negative binomial regression model that accounts for both the distributional properties of accident count data and heterogeneity. Other studies of this issue have not used either panel data in this way nor have they used appropriate statistical methods for count data. Results vary with the selection of the time series used. Overall results suggest that while there may have been an association between fleet fuel efficiency improvements and traffic fatalities in the 1970s, this has largely disappeared. There are suggestions that variance in the composition of the vehicle fleet may have adverse safety impacts.