An analysis of how various road infrastructure improvements affect traffic-related fatalities and injuries is conducted while controlling for other factors known to affect overall safety. The road infrastructure elements analysed include total lane miles, the fraction of lane miles in different road categories (interstates, arterial, and collector roads), the average number of lanes for each road category, and lane widths for arterials and collector roads. Other variables that are controlled for in the study include total population, population age cohorts, per capita income, per capita alcohol consumption, seat-belt legislation (and seat-belt usage), and a proxy variable that represents underlying changes in medical technology. The data used is a cross-sectional time-series database of US states and is analysed using a fixed effects negative binomial regression that accounts for heterogeneity in the data. Data from all 50 states over 14 years is used. Results strongly refute the hypothesis that infrastructure improvements have been effective at reducing total fatalities and injuries. While controlling for other effects, it is found that demographic changes in age cohorts, increased seat-belt use, reduced alcohol consumption and increases in medical technology have accounted for a large share of overall reductions in fatalities.