In order to investigate the damage to microstructure and some other micromechanical responses during a fatigue test on asphalt mixture, Particle Flow Code (PFC) was used to reconstruct a two-dimensional discrete element model of asphalt mixture, based on computed tomography (CT) images and image-processing techniques. The indirect tensile fatigue test of asphalt mixture was simulated with this image-based microstructural model, and verified in the laboratory. It was found that there were four stages during the fatigue failure: no crack, crack initiation, crack developing, and interconnected crack. Cracks mainly developed between the aggregate and asphalt mortar, near the loading axis. The corresponding stages of failure, the developing trend and the distribution characteristics of the cracks matched well with those in the laboratory test. Furthermore, the trends of both the time-load curve and time-displacement curve from the simulation test were also consistent with those from the experimental test. In short, the distribution characteristics of cracks and internal forces of asphalt mixture show that it is feasible to simulate the fatigue performance of the asphalt mixture by a discrete element method (DEM). Appl. Sci. 2019, 9, 327 2 of 17 by the DEM [7]. A local discrete element model was established to discuss the damage and cracks caused by the cyclic fatigue load [8]. Yu analyzed the effect of the aggregate packing on the dynamic modulus of an asphalt mixture by the DEM [9]. These achievements mentioned above demonstrate that the DEM is applicable for the research on fatigue characteristics of pavement asphalt mixes. However, the discrete element models of asphalt mixtures are generated by computer algorithms, which are different from the actual microstructures. The distribution characteristics of cracks and internal forces of asphalt mixtures are also seldom described in the indirect tensile fatigue test.Since microstructure images captured by X-ray computed tomography (X-ray CT) can show the internal distribution and composition of the material, CT and the DEM have been widely applied to asphalt mixture together recently. You et al. obtained the images of asphalt mixture samples by the optical scanning, and reconstructed two-dimensional discrete element models based on these optical images [10]. Based on many two-dimensional slices of the same sample, the three-dimensional discrete element model of an asphalt mixture was reconstructed [11,12]. Zelelew et al. described the microstructure of asphalt mixtures obtained by CT and established a two-dimensional discrete element model to predict the creep compliance of uniaxial loading [13]. Zhou investigated the dynamic deformation process of an asphalt mixture based on the image-based discrete element model [14]. Tan believed that image-based discrete element model could provide a potential detailed insight into the failure mechanism of the heterogeneous rocks at the microscopic level [15]. Generally speaking, the asphalt mixture is a non-uniform complex visco...