Abstract Numerous seafloor observations show that the sizes of high-temperature hydrothermal sulfide edifices vary dramatically with spreading rate. On fast-spreading ridges venting occurs from spindly chimneys which reach heights of about 10 m, while on slow-spreading ridges vents are frequently located near the tops of large sulfide mounds whose volumes may reach 10 5–10 6 m 3. We argue that such variations are the result of a profound difference in the nature of hydrothermal heat extraction between fast-spreading ridges, where spreading occurs predominantly by magmatism, and slow-spreading ridges, where there is a significant component of tectonic extension. Along fast-spreading ridges, a steady-state axial magma chamber can be insulated by a relatively thick conductive boundary layer because heat extraction is limited by low permeability. In such systems, episodes of vigorous venting are linked to diking events not only because these introduce a significant heat source into the upper crust but also because the intrusion of a dike is the primary source of increased permeability near the ridge axis. Over a time scale of the order of a decade, mineral deposition tends to clog the reaction and upflow zones and there is a high probability that young sulfide structures will be buried by subsequent eruptions. On slow-spreading ridges, tectonic extension maintains the fluid pathways necessary to support vigorous convection. In the waning stages of magmatism, hydrothermal circulation driven by a downward-migrating cracking front can cool the entire crust leading to the formation of very large sulfide deposits. As the depth of circulation increases, overburden pressures reduce the permeability and the hydrothermal heat fluxes decrease progressively. Once hydrothermal fluids penetrate the Moho, serpentinization clogs the fluid pathways and high-temperature venting ceases until it is reactivated by a fresh magmatic intrusion.