Two key questions are raised by the presence in the Basin and Range Province of subhorizontal normal faults, which were active in and around the Miocene. First, how did they develop, when their existence seems to be precluded by elementary rock-mechanics arguments? Second, if they did slip in the recent geological past, why do normal faults now active in the brittle upper crust have steeper dips, both in the western USA and elsewhere? One may now answer both questions, by considering the stress field in regions where the isostatic response during extension involves horizontal flow in the plastic lower crust. This stress tensor will have inclined principal axes, making it possible for normal faults to form with either very steep or subhorizontal dips depending on the sense of flow. This can be controlled by environmental conditions, which influence rates of hanging-wall sedimentation and footwall erosion and thus determine lateral variations in lithostatic pressure in the lower crust. Regional-scale lateral variations in the geothermal gradient caused by subduction can also influence lower-crustal flow regimes in adjacent zones of continental extension, enabling extension to occur on low-angle normal faults dipping away from the active continental margin.