Historically, anthropometric test devices (ATDs) designed to predict the risk of thoracic skeletal injuries in frontal crashes have been limited in their ability to discriminate injury independent of the restraint environment. Efforts to improve the accuracy of the ATD-based injury risk functions include the consideration of multi-point thoracic deflection information capable of characterizing the overall deformation of the rib cage independent of the applicable restraint type. The most advanced ATDs in terms of thoracic instrumentation measure multi-axis deflection at four locations, right and left aspects on the 4th and the 8th ribs. The objective of the current study is to assess whether thoracic deflection as measured at four locations provide sufficient information to reliably predict rib injury risk for a range of realistic restraint loading patterns. The computational methodology used to answer the above research question first evaluates the ability of a human finite element (FE) torso model to reproduce regional stiffness characteristics of the thorax as observed in restraint tests involving post mortem test subjects (PMHS). The torso model was then dynamically loaded to a nominally injurious level of deflection by four restraint types (i.e., diagonal belt, distributed load impactor, steering wheel hub impactor and cross-over dual belt) to estimate the true strain distribution within the ribs for each restraint condition. Following the restraint simulations, the thoracic model was deformed using the four-point deflection information measured in the first simulation as the input. The error in peak strain estimation was determined between the true strain distribution resulting from the restraint induced deformation of the thorax and the matched simulation where the displacements at the four thoracic locations were used to deform the thorax. The simulation results indicated that although the torso model has been previously validated for overall restraint loading response the ability of the model to characterize regional loading response was comparatively less biofidelic. Consequently, the error in estimating the peak thoracic strain ranged from 9% to 44% depending on the restraint type;​ however, the location of the true peak strain was accurately predicted in the simulations with four point deflection information. It is important to note that the deformation of the thorax using four point deflections is an extreme case of localized point loading and the estimated strain distribution is thereby susceptible to artifactual errors. While the results from the study reports on the strain error estimates for the worst-case scenario, a more realistic evaluation of the ATD’s ability to predict rib injury risk may require thoracic boundary conditions to be more representative of the ATD characteristics including structural stiffness and material properties.