A finite-difference, thermal balance model of hydrothermal circulation in a sedimentary basin is used to test quantitatively whether large-sized lead-zinc deposits can result from a system such as that postulated by Russell (1978), consisting of a one-pass convection cell within fractures that advance downward into basement rocks. The only heat source is the conductive geothermal flux, and the fluid is a single hypersaline brine recharged from an overlying sea.The buoyancy-driven flow system consists of downward percolation throughout the whole region, which is modeled as a porous medium, and rapid, localized upflow, which is modeled as turbulent pipe flow. The upflow may be located anywhere within the basin where there is a zone of high permeability. A flow equation is derived which equates the buoyancy force to the frictional losses in the recharge and discharge. Heat exchange is calculated between fluid and rock within the porous medium, which is treated as a one-dimensional vertical array of horizontal slices.Downward penetration of the fault system is controlled by an arbitrary cracking temperature, so that the fluid temperature cannot rise above a maximum value (200 degrees -250 degrees C) based on fluid inclusion data.The model determines the mass flow rate and fluid temperature. From this, the mass of Pb + Zn + Fe sulfide which the system could supply is estimated, using a metal solubility-temperature relationship derived from brine analyses.Results show that this system is capable of forming the very largest Pb-Zn deposits, provided that the metals are available to the system. For example, nearly 20 million metric tons of sulfide could be generated in 50,000 years either by circulation penetrating to a 10-km depth in rocks with a geothermal gradient of 30 degrees C km (super -1) or by circulation to a 6-km depth if the geothermal gradient is 60 degrees C km (super -1) . The model was extended to simulate repeated pulses of circulation and showed that thermal reequilibration of the rock to near the ambient gradient takes on the order of 10 6 years after which time heat extraction can be repeated.