The only standardized test procedure for Vehicle rollover tests is described in FMVSS 208. An often raised objection to this test procedure stresses the fact that, for a given car, the number of revolutions as well as the final position of the test car after the crash vary from test to test. Consequently, there are no body deformations that resemble those of the previous test.

This paper presents an analysis of the contact forces. Such Contact forces are determined from the accelerations of the car bodies during the rollover employing a numerical calculation method. It became obvious that the structural forces are reproducible, in spite of the differences in rollover kinematics. The body load strongly depends on the geometric design and the stiffness distribution of the car body. Another conclusion of the test series is that good results in roof crush tests are not necessarily an indicator for good rollover performance.

Biomechanical data are measured by Hybrid III dummies. In addition to measuring the forces and torque of the head and the neck, the surface pressure on the skull was determined by using a pressure sensitive foil. With the exception of extreme roof crushes, there is no correlation between roof crush and biomechanical loads. This result is in accordance with the analysis of real world crashes.

From the test series one may conclude that rollover crashes are a useful instrument for developing cars. However, such crashes only allow a qualitative investigation of crash formance, but they are not suitable for establishing any ratings of safety for different types of cars.