Two candidate functions for the prediction of pelvic fracture probability are created using lateral pelvic acceleration and age data from 84 cadaver impact experiments. These functions include measures of bone stress, bone strain, age, and load concentration factor. Data are analyzed using the maximum likelihood approach, and it Is found that representation of the data by a Weibull distribution yields higher maximum likelihood than does the assumption of an underlying normal distribution.

Additional data are presented documenting pelvic force and acceleration responses for both cadavers and the Side Impact Dummy (SID). Systematic differences are noted and discussed.