This paper is included in the Special Publication entitled 'Faulting, fault sealing and fluid flow in hydrocarbon reservoirs', edited by G.Jones, Q.J. Fisher and R.J. Knipe. Two-dimensional, porous-medium, steady-state, coupled fluid- and heat-transport models are used to investigate some of the hydrogeological and thermal consequences of steeply dipping fault damage zones in a normally pressured basin setting. Simple geometries and conservative petrophysical properties can produce large-scale buoyancy-driven circulation, both outside and within the fault zone. An average (homogeneous) basin permeability of only 7 mDk(h) (horizontal permeability) and 0.07 mD k(v) (vertical permeability) results in a free convection cell of this type, with the upflow being localized by a 300 m wide fault zone (50 mD k(h), 0.5 mD k(v)). Steady-state temperature anomalies as large as 15°C at the top of the fault zone can be produced by this arrangement. Smaller values of basin permeability still result in a similar circulation pattern, but at flow rates which produce temperature anomalies that are below detection levels. When the basin fill is more heterogeneous (layered), higher permeabilities can exist in some layers without large-scale convection occurring, because of the dampening effect of other, lower permeability units. In realistic geometrical configurations that are similar to the North Sea Central Graben, the fluid flow system is dominated by within-fault-zone convective circulation that produces local (<10 km half-wavelength), high amplitude (50°C) temperature anomalies which are comparable to the largest of those actually observed in the subsurface.