Different types of hydraulic tests (pulse, recovery, cross-hole, and tunnel inflow measurements) have been performed in a low-permeability fractured granite around the Full-scale Engineered Barrier Experiment (FEBEX) gallery in Grimsel (Switzerland). We have interpreted the tests using conventional methods that treat the medium as a homogeneous one. Results display scale effects. Hydraulic conductivities increase, by orders of magnitude, with the volume of rock tested (from pulse to cross-hole tests). The objective of our work is to show that this scale effect is apparent. It reflects the limitations of the equivalent hydraulic conductivity derived from the homogeneous model interpretation of the tests. For this purpose we built a three-dimensional model for the joint interpretation of cross-hole tests. The model represents explicitly the conductive fractures and treats the rest as an equivalent porous medium (matrix). Transmissivity values used in the model to represent fractures are consistent with those derived from cross-hole tests and the few single-hole tests at intervals intersecting fractures. On the other hand, matrix hydraulic conductivities are consistent with the remaining single-hole, short-time tests. This model can also be used to simulate observed heads and gallery inflows. In summary, the final model is consistent with all the relevant measurements, taken at different support scales. This provides some insight into the issue of scale effects, which has been a topic of debate in the literature. In summary, the majority of small-scale tests are performed in matrix intervals. Thus any averaging of these values would suggest relatively small effective permeability. Yet large-scale permeability of the rock is controlled by a few fractures, which provide high connectivity to the system but are intersected by few testing intervals. As a result, large-scale permeability is qualitatively different and quantitatively larger than small-scale permeability.