High-precision Mg isotope measurements by multiple collector inductively coupled plasma mass spectrometry were applied for determinations of magnesium isotopic fractionation of biogenic calcium carbonates from seawater with a rapid Mg purification technique. The mean δ(26)Mg values of scleractinian corals, giant clam, benthic foraminifera, and calcite deep-sea corals were -0.87‰, -2.57‰, -2.34‰, and -2.43‰, suggesting preferential precipitation of light Mg isotopes to produce carbonate skeleton in biomineralization. Mg isotope fractionation in deep-sea coral, which has high Mg calcite skeleton, showed a clear temperature (T) dependence from 2.5 °C to 19.5 °C: 1,000 × ln(α) = -2.63 (±0.076) + 0.0138 (±0.0051) × T(R(2) = 0.82, p < 0.01). The δ(26)Mg values of large benthic foraminifera, which are also composed of a high-Mg calcite skeleton, can be plotted on the same regression line as that for deep-sea coral. Since the precipitation rates of deep-sea coral and benthic foraminifera are several orders of magnitude different, the results suggest that kinetic isotope fractionation may not be a major controlling factor for high-Mg calcite. The Mg isotope fractionation factors and the slope of temperature dependence from deep-sea corals and benthic foraminifera are similar to that for an inorganically precipitated calcite speleothem. Taking into account element partitioning and the calcification rate of biogenic CaCO(3), the similarity among inorganic minerals, deep-sea corals, and benthic foraminiferas may indicate a strong mineralogical control on Mg isotope fractionation for high-Mg calcite. On the other hand, δ(26)Mg in hermatypic corals composed of aragonite has been comparable with previous data on biogenic aragonite of coral, sclerosponges, and scaphopad, regardless of species differences of samples.