Red light running is a significant public health concern, killing more than 800 people and injuring 200,000 in the United States per year (Retting et al., 1999a; Retting and Kyrychenko, 2002). To reduce red light running in Virginia, six jurisdictions (Alexandria, Arlington, Fairfax City, Fairfax County, Falls Church, Vienna) deployed red light cameras at some point during the 10-year period when they were permitted under Virginia law. This report documents the safety impacts of those cameras based on 7 years of crash data for the period January 1, 1998, through December 31, 2004. Consistent with the findings of a previous Virginia study (Garber et al., 2005), this study finds that cameras are associated with an increase in rear-end crashes (about 27% or 42% depending on the statistical method used as shown in Tables ES1 and H1) and a decrease in red light running crashes (about 8% or 42% depending on the statistical method used as shown in Tables ES1 and H2). This report also shows that there is significant variation by intersection and by jurisdiction: one jurisdiction (Arlington) suggests that cameras are associated with an increase in all six crash types that were explicitly studied (rear-end, angle, red light running, injury red light running, total injury, and total) whereas two other jurisdictions saw decreases in most of these crash types. It is therefore not surprising that when the comprehensive crash costs for rear-end and angle crashes are monetized, the cameras are associated with an increase in crash costs in some jurisdictions (e.g., an annual increase of $140,883 in Arlington) and a net reduction in comprehensive crash costs in other jurisdictions (e.g., an annual reduction of $92,367 in Vienna). When these results are aggregated across all six jurisdictions, the cameras are associated with a net increase in comprehensive crash costs. However, when considering only injury crashes, if the three fatal angle crashes that occurred during the after period are removed from the analysis (the only fatalities that occurred during the study out of 1,168 injury crashes), then the cameras were associated with a modest reduction in the comprehensive crash cost for injury crashes only. These results cannot be used to justify the widespread installation of cameras because they are not universally effective. These results also cannot be used to justify the abolition of cameras, as they have had a positive impact at some intersections and in some jurisdictions. The report recommends, therefore, that the decision to install a red light camera be made on an intersection-by-intersection basis. In addition, it is recommended that a carefully controlled experiment be conducted to examine further the impact of red light programs on safety and to determine how an increase in rear-end crashes can be avoided at specific intersections.