Hydraulic fracturing in rock masses involves complex and coupled processes of fractures propagating in discontinuous media and of fluid flow in discrete channels. The Unconventional Gas Program at Lawrence Livermore National Laboratory is actively developing numerical models to gain insight in the physics of these coupled processes. However powerful and versatile these models are, they must be verified against analytical solutions and/or physical experiments. This paper describes the current status of the FEFFLAP code, which models fluid-driven discrete fracture propagation in jointed media. We also discuss the results of a preliminary series of physical tests in which hydrofractures were driven across slanted interfaces between dissimilar materials, in blocks loaded in biaxial compression.