We present a retrospective evaluation of GHBMC human body models related to thoracic injury prediction. This post-hoc analysis matched post-mortem human surrogates (PMHS) rib fracture outcomes to GHBMC simulations and compared the number of ribs with one or more fractures, number of fractured ribs(NFR). Thirty simulations were analysed with the following models:​ F05-O v6.0, F50-O v6.0.1, M50-O v6.1, F05-70YO v6.0, M50-70YO v6.0. NFR was used for consistent analysis between literature and GHBMC simulations. Load cases included hub, sled, and airbag loading in frontal, lateral, rear, and oblique directions. Rib fracture was assessed probabilistically using the Larsson 2021 risk curve and average PMHS age, unless an age targeted model was used, in which case the age was set to the model age. Fracture prediction is also evaluated using a recently published GHBMC-specific risk curve by Forman. GHBMC models on aggregate predicted 70% of PMHS fractured ribs with an R2 of the linear correlation between model and PMHS being 0.55. Hub and airbag loading cases showed strong correlation whereas more complex simulations showed less. It is unclear if the findings indicate an under prediction of fracture by the model, or reflect known limitations of PMHS such as greater likelihood of osteopenia. This retrospective study reflects the performance of GHBMC HBMs from a broader view than typically presented in a single study. Aggregation by families of models in test types and loading conditions provide a means to identify potential areas of improvement for the model.