This paper aims to investigate the influence of void characteristics on the uniaxial penetration strength (UPS) of asphalt mixtures using a three-dimensional (3D) discrete-element method (DEM). Three models for the uniaxial penetration test (UPT) for asphalt mixtures (namely, a model considering real void size and distribution, a model considering random void size and distribution, and a model without considering voids) were created using imaging techniques and 3D DEM embedded with a Burgers model. The UPS was predicted/simulated on the basis of the models. The simulation results were verified against known laboratory UPTs. Research results showed that the UPS of asphalt mixtures was able to be predicted by using 3D DEM, and void characteristics had a significant influence on the UPS. With the same model parameters, the UPS of the model without considering voids was the largest, followed by that of the model considering random void size and distribution, whereas the model considering real void size and distribution was the lowest but the closest to the experiment results. Furthermore, aggregate size, binder content, and temperature affected the UPS. The UPS grew with increasing nominal maximum aggregate size and declined with temperature. The UPS with low or high binder content was less than that with the optimum binder content at 60°C.