Seafloor hydrothermal systems clearly have played an important role in controlling the composition of seawater over geologic time. However, controversy exists concerning their role as drivers of secular variability in major element composition of seawater during the Phanerozoic. The history of Phanerozoic changes in calcium (Ca2+) and magnesium (Mg2+) concentration has been reconstructed through analysis of fluid inclusions in evaporites, and this history is inconsistent with global geochemical cycling models that link both volcanic CO2 release and Mg2+ uptake/ Ca2+ release to inferred rates of seafloor production. These models generate little variability in Mg2+ concentrations because of compensating effects. Some models that recreate the observed trends either do not conserve alkalinity or do not link seafloor spreading, inferred from past variations in continental flooding presumed to reflect changing mid-ocean-ridge volumes, to volcanic CO2 production. The most comprehensive models reproduce the variations in seawater Mg2+/Ca2+ through time quite well, but not because this ratio is tied to variations in seafloor spreading rate, and despite considerable mismatches to the Mg2+ record. This cacophony arises in part from our lack of quantitative understanding of fundamental relationships between heat flow, seafloor production rates, sea level, hydrothermal circulation rates, volcanic CO2 release rates, and ocean chemical changes. Nevertheless, the current fluid inclusion data seem to indicate that ocean composition has an “attractor” that drives Ca2+ and Mg2+ toward equilibration with seafloor hydrothermal mineral assemblages [20 mm (millimolal) Ca2+ and 0 mm Mg2+], thwarted by other processes (weathering and riverine inputs, biogenic mineral precipitation) that drive the system away from this attractor.