Though many studies on the Mg contents in the calcitic tests of foraminifers exist, the processes controlling its uptake are still a matter of debate. Laboratory cultures offer an excellent opportunity to reveal these mechanisms. The Mg concentrations within single chambers of the planktic foraminifer Globigerinoides sacculifer (BRADY) maintained under controlled laboratory conditions were measured (1) at variable temperatures (19.5-29.5°C) and constant salinity and (2) at variable salinity (22-45‰) and constant temperature. The experimental results suggest that under natural conditions, temperature is the leading mechanism controlling the Mg/Ca ratio. Temperature and magnesium are related proportionally. A temperature increase of ca. 10°C gives rise to an increase of the magnesium concentrations of ca. 130%. Drastic (unnatural) salinity changes dominate the effects of temperature. A 110% change in the Mg/Ca ratio was observed when salinity was elevated or reduced by more than ca. 10‰. Specimens which underwent gametogenesis reveal significantly higher Mg concentrations than specimens that did not release gametes. Partition coefficients for Mg in foraminiferal calcite are orders of magnitude lower than values from inorganically precipitated calcite. When comparing observed Mg/Ca ratios of foraminiferal tests with predicted Mg/Ca ratios calculated according to empirical equations, it becomes evident that foraminiferal tests are undersaturated with respect to Mg for the water temperature they have experienced. Apparently, foraminifers are capable of controlling their Mg concentration. The physiological processes presumably responsible for such depressed Mg/Ca ratios appear to be temperature-controlled as deduced from the close relationship of the observed Mg/Ca ratios and water temperature. This study demonstrates that variations in temperature and salinity are definitely reflected in the Mg content of foraminiferal tests. Magnesium may thus serve as a paleo-proxy for past surface water temperatures, as long as postdepositional changes and salinity variations are of subordinate importance or can be excluded.