Binary granular mixtures are comprehensively explored from both macro- and microscopic perspectives. In this study, the undrained behavior of binary granular mixtures constituted by fine and coarse particles is investigated using the discrete element method (DEM) under different fines contents (FCs). The peak deviatoric stress ratio has a linear relation with the effective void ratio. The increase in the FC to a certain value (approximately 10% in this work) can change the material response from strain hardening to limited "liquefaction". The enhanced strain hardening behaviors were observed when the FC exceeds 15%. The location of the critical state line (CSL) depends on the fines content. Voronoi tessellation was adopted to explore the mesoscale structures of binary mixtures. The local isotropy, local anisotropy, and sphericity of the Voronoi cells are clearly related to the FC. The radial distribution functions (RDFs) also exhibit clear correlations with the FC. The abundance and mechanical contributions of different types of contacts, e.g., coarse-coarse, coarse-fine, and fine-fine contacts, are analyzed. For these binary systems, the proportions of the sliding contacts with respect to the FCs are contact type dependent. The coaxiality between the loading direction and fabric orientation of the different contact types depends on the relative proportions of the contact types. The evolutions of the global anisotropies are also explored, and the contact-based anisotropy parameters extracted at peak states under various FCs are compared to the global anisotropy parameters. The stress-force-fabric relationship is evaluated in various subnetworks of the mixtures and discussed in terms of the strong-weak partition and nonsliding-sliding partition networks. The dominant role of geometrical anisotropy derived from the strong subnetwork is emphasized, and the linear relationship between the global stress ratio and the geometrical anisotropy within strong-nonsliding subnetwork is evidently observed under various conditions of fines content. (C) 2018 Elsevier B.V. All rights reserved.