Am, a Dodge Coronet, an American Motors Gremlin, a Chevrolet Brookwood station wagon, and an Oldsmobile Toronado.) Four primary control input patterns, includ ing reverse steer, reverse steer and braking, a slalom maneuver, and a trapezoidal steer input, were varied in magnitude and timing. Variations on these basic maneu vers included steering actions of fixed, free, and return-to-zero operation, test speeds ranging from 25 to 45 mph, constant and closed throttle control, a range of steering input rates, and combinations of inputs. On board instrumentation measured speed, steering input angle, lateral acceleration, roll angle, roll angular rate, and yaw rate. The authors performed over 500 test runs varying the configuration and oper ation of each of the vehicles. The only vehicle which rolled over was the 1974 Ford Pinto, with its shock absorbers removed and with added roof weight, in the simple reverse steer maneuver in the speed range of 33-35mph. (This vehicle was used in more than 200 test runs, and later this rollover response could not be reproduced.) The authors stated that they could not devise a general test procedure for producing vehicle rollover under realistic flat-surface conditions. They found that they could not determine one set of operating values for speed, control input characteristics, etc., which would make all vehicle configurations roll over. But they did conclude that T/2h is closely related to vehicle rollover resistance (T is the vehicle's track width and h is the center of gravity height). Furthermore, they concluded that roll resis tance can be degraded by design and operational features such as worn suspension elements (shock absorbers), short travel distance for compression bump stop contact, and spiked application of brakes at conditions of maximum cornering. Rice et al. also simulated rollover using the Highway-Vehicle-Simulation Model