This paper attempts to define and measure factors related to a vehicle's performance that are influential in the causation of rollover accidents. Data are presented which define the rollover involvement rates for many non-vehicular factors. A brief description of the vehicle metrics and the analysis procedures used in the rollover prevention rulemaking program are included along with a set of conclusions.The program evaluated many vehicle metrics related to vehicle rollover, analyzed accidents from 5 states, and compared the two data bases by testing “cause and effect” hypotheses by performing statistical regressions to determine levels of correlation. Location of the crash, urban vs. rural, was a strong predictor of the crash outcome - that is, rollover or non-rollover. Vehicle class and single vehicle accident rate were also statistically significant, as well as, whether or not the vehicle was equipped with anti-lock brakes. Several other driver demographics were significant. A good relationship was found between one of the vehicle rollover metrics and the likelihood of rollover given that an accident has occurred. Additional analyses are being conducted to better understand the effects of directional stability vehicle metrics in the prediction of rollover crashes.