Enamel, the hard surface layer of teeth, is a three-dimensional biological composite made of crisscrossing mineral rods bonded by softer proteins. Structure-property relationships in this complex material have been difficult to capture and usually require computationally expensive models. Here we present more efficient discrete element models (DEM) of tooth enamel that can capture the effects of rod decussation and rod-to-interface stiffness contrast on modulus, hardness, and fracture resistance. Enamel-like microstructures were generated using an idealized biological growth model that captures rod decussation. The orthotropic elastic moduli were modeled with a unit cell, and surface hardness was captured with virtual indentation test. Macroscopic crack growth was also modeled directly through rupture of interfaces and rods in a virtual fracture specimen with an initial notch. We show that the resistance curves increase indefinitely when rod fracture is avoided, with the inelastic region, crack branching, and 3D tortuosity being the main sources of toughness. Increasing the decussation angle simultaneously increases the size of the inelastic region and the crack resistance while decreasing the enamel axial modulus, hardness, and rod stress. In addition, larger contrasts of stiffness between the rods and their interfaces promote high overall stiffness, hardness, and crack resistance. These insights provide better guidelines for reconstructive dental materials, and for development of bioinspired hard materials with unique combinations of mechanical properties. Statement of SignificanceEnamel is the hardest, most mineralized material in the human body with a complex 3D micro-architecture consisting of crisscrossing mineral rods bonded by softer proteins. Like many hard biological composites, enamel displays an attractive combination of toughness, hardness, and stiffness, owing to its unique microstructure. However few numerical models explore the enamel structure-property relations, as modeling large volumes of this complex microstructure presents computational bottlenecks. In this study, we present a computationally efficient Discrete-element method (DEM) based approach that captures the effect of rod crisscrossing and stiffness mismatch on the enamel hardness, stiffness, and toughness. The models offer new insight into the micromechanics of enamel that could improve design guidelines for reconstructive dental materials and bioinspired composites.