The paired cranial crests of Sinosaurus(Theropoda) have been hypothesized as too weak to resist mechanical loads during combat. Finite element analysis(FEA) is used to test this hypothesis, first with geometry obtained through direct laser scanning of a well-preserved fossil of the crest, and then with two conceptual FE models of both crests analyzing the structure-deformation effects of fenestration. In the original fossil model, under direct loading on the dorsal faces of the crest, we found that the areas surrounding cavities on the crest experience shear stress that implies a high chance of material failure – the fracture of bone. In the conceptual model, a series of computational studies were conducted with varying loading directions. One simulation found that the shear stress and strain in the material around the cavity presented more deformation compared with the conceptual model without the cavities, and under this morphologically realistic scenario the loading conditions would result in local bone fractures. These model-based computational results indicate that the crest could not resist high loads, because it could not effectively decentralize the loading stress. Future investigations need to focus on more comprehensive computational experiments with more conditions, e.g. dynamical loading conditions, and direct palaeontological evidence. The paired cranial crests of Sinosaurus (Theropoda) have been hypothesized as too weak to resist mechanical loads during combat. Finite element analysis (FEA) is used to test this hypothesis, first with geometry obtained by direct laser scanning of a well-preserved fossil of the crest, and then with two conceptual FE models of both crests analyzing the structure-deformation effects of fenestration. In the original fossil model, under direct loading on the dorsal faces of the crest, we found that the surrounding cavities on the crest experience shear stress that implies a high chance of material failure - the fracture of bone. In the conceptual model, a series of computational studies were conducted with varying loading directions. One simulation found that the shear stress and strain in the material around the cavity presented more deformation compared with the conceptual model without the cavities, and under this morphologically realistic scenario the loading conditions wou ld result in local bone fractures. These model-based computational results indicate that the crest could not resist high overload, because it could not effectively decentralize the loading stress. Future investigations need to focus on more comprehensive computational experiments with more conditions, eg dynamical loading conditions, and direct palaeontological evidence.