Various mechanisms have been proposed to explain the salinity variations in vent fluids of seafloor geothermal systems. New experiments reacting diabase and evolved seawater were carried out to reproduce earlier published observations of Cl depletions attributed to formation of an ephemeral Cl-bearing mineral. The absence of any Cl depletions in the present study suggests that the formation of Cl-bearing minerals is not sufficiently widespread to account for the observed salinity variations in the vent fluids. A re-evaluation of both field and laboratory evidence has led to a new model for subseafloor circulation that accounts for salinity variations as well as other chemical and mineralogic observations. In place of a simple single-pass convection system, we propose that the seafloor systems consist of two vertically nested convection cells in which a brine layer at depth heats and drives an overlying seawater cell. Such layering of salinities, a process known in fluid mechanics as double-diffusive convection, is an expected result when convection is induced in saline fluids. The process provides for stable high-temperature heat transfer upward from the cracking front adjacent to the magma, and for limited chemical exchange of the brine with the overlying seawater to explain salinity variations and high metal contents in the vent fluids. The brine also provides an effective medium to produce the secondary mineral assemblages observed in rocks from the mid-ocean ridges and ophiolites unsuccessfully produced in laboratory studies using seawater. The brine originates from the two-phase separation of seawater during magmatic/tectonic events and accumulates and remains relatively stable in the region immediately above the magma chamber.