A series of rollover tests was conducted in a real-world environment in which a vehicle was driven or towed to highway speed then steered to induce a rollover. This research presents analysis of the rollover phase of five tests. In each test, the steering maneuver was initiated on-pavement, and the rollover was caused by tire-to-ground interaction. Tests included vehicles that tripped both on-pavement and on soil. Four tests ended with the vehicle at rest off-road, and one ended with the vehicle remaining on the pavement. A programmable remote control radio was used to steer the vehicles through a double-step steer maneuver to result in a rollover. The test vehicles were instrumented and data was collected during each test, including steering, suspension motion, rotational rates, and accelerations. A Global Positioning System (GPS) speed sensor (VBOX III manufactured by Racelogic) was used to monitor the vehicle speed. Data from all tests is presented in the Appendix . Each test was recorded from multiple perspectives using real-time and high-speed video cameras. The electronic data acquisition was synchronized with the vehicle's roadway position and with each video camera using flashbulbs and a chalk gun. The crash site evidence was documented through surveying and photography. The scratch patterns and crush damage to each test vehicle were studied and photographed. Each test was analyzed in detail, including calculation of the speed over the ground throughout the rollover, drag factors, and rollover distances. The calculated speed at trip was compared to the output of the VBOX speed sensor. Based on this comparison, validation testing of the VBOX was conducted and a correction formula for VBOX trip speed is proposed.