Willemite (zinc silicate) is the main zinc mineral in some carbonate-hosted ore deposits (e.g., Franklin, New Jersey; Vazante, Brazil; Beltana, South Australia; Kabwe, Zambia). Recent interest in these unconventional zinc deposits has increased because of high zinc grades that exceed 40 wt percent, relatively low environmental impact of ore processing owing to the lack of acid-generating sulfides in the waste, and advances in ore processing technologies. In the past, most metallogenic studies proposed formation of willemite deposits by supergene or hypogene alteration of preexisting sulfide deposits. However, recent data on the Vazante, Beltana, and Kabwe deposits indicate willemite crystallization at temperatures in excess of 150°C, raising the possibility of primary precipitation from hydrothermal fluids. We use numerical geochemical modeling to examine the formation of willemite under hydrothermal conditions. Activity-activity diagrams reveal that, in the presence of dissolved sulfur and quartz, willemite instead of sphalerite will precipitate under oxidizing (e.g., hematite-stable, sulfate-predominant) and alkaline (pH higher than K feldspar-muscovite-quartz) conditions. Willemite also becomes more stable, relative to sphalerite, at high temperature, and willemite call coexist with magnetite at 300°C. The stabilities and solubilities of sphalerite, willemite, smithsonite, hydrozincite, and zincite were calculated for wide ranges of temperature (25°-300°C), chloride concentration, dissolved sulfur and carbon concentrations, pH, quartz saturation, and oxidation potential. Plots of the solubility of the different minerals as a function of two variables (e.g., temperature and redox state; pH and redox state) allow us to predict the effects of changing chemical conditions, which in turn permits an estimate of the efficiency of particular precipitation processes. Cooling is an effective process for precipitating sphalerite but not willemite, whereas pH increase (e.g., by acidic fluids reacting with carbonates) is effective for precipitating willemite but not sphalerite. Dynamic geochemical models that simulate physicochemical processes are used to understand the formation of the Beltana willemite deposit in the Adelaide geosyncline of South Australia. This small, high grade deposit (850,000 t at 36% Zn) is hosted in dolomite of the Cambrian Ajax Limestone, next to a tectonic contact with the diapiric, halite-bearing clastic sediments of the Callanna Group. The orebody is associated with hematite alteration and is characterized by the total absence of sulfides; willemite is the only zinc ore mineral, and the arsenate hedyphane (Ca 2 Pb 3 [AsO 4 ] 3 Cl) is the main lead mineral. The model results show that willemite will precipitate in response to water-rock interaction and fluid mixing processes at temperatures above 120°C. The presence of arsenate in the hydrothermal fluid is likely to have been important at Beltana; in arsenate-absent models sulfate is reduced to sulfide by the precipitation of ferrous iron as hematite, resulting in the precipitation of sphalerite and galena. In contrast, in models including arsenate the reduction of sulfate to sulfide is inhibited and willemite is predicted to precipitate.