Abstract: The role of polymetallic melts in scavenging ore components has recently been highlighted in the context of fluid-poor metamorphosed ore deposits. In contrast, the role of polymetallic melts in systems dominated by hydrothermal fluids remains poorly understood. Using a simple Au–Bi model system, we explored experimentally whether such polymetallic melts can precipitate directly from a hydrothermal fluid, and investigated the ability of these melts to scavenge Au from the solution. The experiments were conducted in custom-built flow-through reactors, designed to reproduce a hydrothermal system where melt components are dissolved at one stage along the flow path (e.g., Bi was dissolved by placing Bi-minerals along the fluid path), whereas melt precipitation was caused further along the flow path by fluid–rock interaction. Bi-rich melts were readily obtained by reaction with pyrrhotite, graphite or amorphous FeS. When Au was added to the system, Bi–Au melts with compositions consistent with the Au–Bi phase diagram were obtained. In the case of fluid reaction with pyrrhotite, epitaxial replacement of pyrrhotite by magnetite was observed, with textures consistent with an interface-coupled dissolution–reprecipitation reaction (ICDRR). In this case, the metallic melt precipitated as blebs that were localized at the replacement front or within the porous magnetite. Direct fractionation of Bi–Au melts from a hydrothermal fluid, or precipitation of a Bi-melt followed by partitioning of Au from ambient fluid, offer new pathways to the enrichment of minor ore components such as Au, without requiring fluid saturation with respect to a Au mineral. This mechanism can explain the strong geochemical affinity recognized between Au and low-melting point chalcophile elements such as Bi in many gold deposits. Examples of deposits where such a model may be applicable include orogenic gold deposits and gold skarns. Contrary to models involving migration of polymetallic melts to explain element remobilization, only small quantities (ppm) of polymetallic melts are required to affect the Au endowment of a deposit via interaction with a hydrothermal fluid. The experiments also show that micro-environments can play a critical role in controlling melt occurrences. For example, reaction fronts developing via ICDR reactions can promote melt formation as observed during the replacement of pyrrhotite by magnetite. The associated transient porosity creates space for the melt and promotes melt-fluid exchanges whereas the reaction front provides local geochemical conditions favorable to melt precipitation (e.g., reduced, low aH2S(aq), and catalytic surface). [Copyright &y& Elsevier]