Experiments were undertaken to investigate how faults (reverse or normal) in a cover sequence become reactivated when an underlying basement fault is itself reactivated in strike-slip mode. Physical models, properly scaled for gravity, were deformed in a large (2 × 1.5 m) apparatus divided into two halves; one half can be moved laterally or vertically past the other. Models were constructed with a Coulomb layer (sand) overlying a Newtonian ductile layer (silicone). Three different configurations of the basement fault (reverse, normal or vertical) defined three series of experiments; each consisted of 1 brittle model (6 cm of sand) and 2 brittle-ductile models (5 cm of sand overlying 1 cm of silicone, or 4 cm of sand overlying 2 cm of silicone). Each model was deformed in two steps: (1) a pure dip-slip stage, and (2) a reactivation in pure strike-slip mode. In surface views, little reactivation was observed, but in all cases, the deformation imposed during the strike-slip episode was localized by pre-existing dip-slip structures. In cross-sections, however, faults were seen reactivated at depth. On the basis of these experiments we suggest that (1) a previously faulted zone can control the deformation of a strike-slip zone, and (2) reactivation of faults in strike-slip mode can occur at depth, without being visible at the surface. We attribute the higher probability of reactivation at depth to the mechanical properties of the sand, especially cohesion.