The purpose of this study was to use detailed medical information to evaluate thoracic injury of elderly patients in real world frontal crashes. In this study, we used analytic morphomics to predict the effect of torso geometry on thoracic injury. This method, extracts body features from computed tomography (CT) scans of patients in a semi-automated fashion. Thoracic injuries were examined in front row occupants involved in frontal crashes from the International Center for Automotive Medicine (ICAM) database. Among these occupants, two age groups (age < 60yr. [Nonelderly] and age ≧ 60yr. [Elderly]) who suffered severe thoracic injury were analyzed. Regression analyses were conducted to investigate injury outcomes considering variables including those for vehicle, demographics, and morphomics. Compared to the nonelderly group, the elderly group sustained more rib fractures. Logistic regression models were fitted with different configurations of variables predictive of the Maximum Abbreviated Injury Scale of thoracic region (MAISthx 3+). The model developed based solely on vehicle data had an area under the receiver operating characteristic curve (AUC) of 0.60. When demographic data was combined with vehicle data, the model prediction improved to an AUC of 0.70. The AUC associated with vehicle and morphomics data increased to 0.74 and increased again to 0.82 when combining vehicle, demographic, and morphomics variables. The important morphomics variables were rib geometry, bone density, and spin-to-back skin, which represents fat thickness in the posterior trunk. Morphomics variables such as skeletal geometry and fat distribution can be precisely adjusted in a finite element human body model or anthropomorphic testing device to represent occupants of different body shapes and sizes and are thus more valuable in assessing injury during vehicle crashes.