The field of motor vehicle rollover research and testing has been one of multiple and varied approaches, dating back to at least the 1930's. The approach has been as simple as tipping a vehicle over at the top of a steep hill ( Wilson et al., 1972 ), to as complex as releasing a vehicle from an elevated roll spit mounted to the rear of a moving tractor and trailer ( Cooper et al., 2001 and Carter et al., 2002 ). Between these extremes exists numerous other rollover initiation methods, including driving a vehicle into a ramp, sliding a vehicle sideways into soil or a curb-like obstruction, launching or releasing a vehicle from a dolly cart, and remotely steering a moving vehicle into an orientation that will produce a rollover event ( Cooperrider et al., 1990 and Larson et al., 2000 ). Each method has its own unique advantages and limitations, including degrees of initial and early condition repeatability, availability for robust dynamic documentation, and degree of applicability to rollover events in the field. The method of releasing a vehicle from a dolly cart, also referred to as dolly rollover testing, has been used in the automotive industry for over 40 years, and its relevance to actual field rollover events has been debated for nearly as long. Debate has centered around the effect the initial conditions have on the subsequent rollover events. Specifically, whether these initial conditions alter the dynamics of the rolling vehicle in a way that is un-natural or beyond that which can occur in the real world. Until recently, however, there has not been a substantial or meaningful amount of data available to directly analyze and compare the two environments, dolly rollover testing and real-world rollovers. Fortunately, the field of rollover experiments and analytical tools has progressed to the point where comprehensive data now exists with well documented steer-induced rollover demonstrations ( Asay et al., 2009 and Asay et al., 2010 ). A substantial body of dolly rollover testing outside of the standard FMVSS 208 / SAE J2114 protocol is now available. Numerous dolly rollovers have been performed well above 30 mph, many between 35 and 45 mph, and some as high as 55 mph. These tests have also included those executed on dirt rollover surfaces. The body of steer-induced rollover tests has also expanded, and now includes numerous rollover events performed at various speeds, with specific classes of vehicles, and conducted off-road in the dirt adjacent to an actual highway. It is this subset of rollovers, events that occur off-road on dirt, at trip speeds above 30 mph that will be examined in detail in this study. The dolly rollover and steer-induced rollover tests that belong in this subset will be analyzed, and comparisons will be made between the two test methods, focusing on vehicle rollover dynamics. Specific dynamic parameters related to the development and progression of rollover events will be studied and related to the initial conditions of each test. These parameters include: rollover distance, number of rolls or quarter turns, average distance per roll, average translational deceleration, average roll rate, peak roll rate, and the development and progression of the roll rate history.