This paper presents a combined laboratory and numerical investigation on the injection-induced permeability changes in pre-existing fractures. The analyses conducted were primarily based on the results of an innovative laboratory experiment designed to replicate the key mechanisms that occur during hydraulic stimulation of naturally fractured rocks and/or faulted zones. The experiment involved pressure-controlled fluid injection into a laboratory-scale pre-existing fracture within a granite block, which was subjected to true triaxial stress conditions. Rough and smooth fractures are investigated, and the results are discussed. Based on the experimental results, two contributing mechanisms were considered to describe the pressure-driven permeability changes in pre-existing fractures: (1) elastic opening/closure leading to a reversible permeability change, and (2) fracture sliding in shear mode, causing dilation and hence an irreversible permeability increase. With these assumptions, an aperture-dependent permeability function was adopted to couple the hydraulic flow with the mechanical deformations along the fracture. Subsequently, a 3D coupled hydro-mechanical model was developed to replicate fluid-injection tests conducted at various conditions, including different stress conditions and fracture surface roughness. The employed modeling framework effectively captured the experimental observations. Our results indicate that the maximum permeability increases twofold.