This survey of magnesium stable isotope compositions in marine biogenic aragonite and calcite includes samples from corals, sclerosponges, benthic porcelaneous and planktonic perforate foraminifera, coccolith oozes, red algae, and an echinoid and brachiopod test. The analyses were carried out using MC-ICP-MS with an external repeatability of + or -0.22 ppm (2SD for delta (super 26) Mg; n=37), obtained from a coral reference sample (JCp-1). Magnesium isotope fractionation in calcitic corals and sclerosponges agrees with published data for calcitic speleothems with an average Delta (super 26) Mg (sub calcite-seawater) =-2.6+ or -0.3 ppm that appears to be weakly related to temperature. With one exception (Vaceletia spp.), aragonitic corals and sclerosponges also display uniform Mg isotope fractionations relative to seawater with Delta (super 26) Mg (sub biogenic aragonite-seawater) =-0.9+ or -0.2. Magnesium isotopes in high-Mg calcites from red algae, echinoids and perhaps some porcelaneous foraminifera as well as in all low-Mg calcites (perforate foraminifera, coccoliths and brachiopods) display significant biological influences. For planktonic foraminifera, the Mg isotope data is consistent with the fixation of Mg by organic material under equilibrium conditions, but appears to be inconsistent with Mg removal from vacuoles. Our preferred model, however, suggests that planktonic foraminifera synthesize biomolecules that increase the energetic barrier for Mg incorporation. In this model, the need to remove large quantities of Mg from vacuole solutions is avoided. For the high-Mg calcites from echinoids, the precipitation of amorphous calcium carbonate may be responsible for their weaker Mg isotope fractionation. Disregarding superimposed biological effects, it appears that cation light isotope enrichments in CaCO (sub 3) principally result from a chemical kinetic isotope effect, related to the incorporation of cations at kink sites. In this model, the systematics of cation isotope fractionations in CaCO (sub 3) relate to the activation energy required for cation incorporation, which probably reflects the dehydration of the cation and the crystal surface and bond formation at the incorporation site. This kinetic incorporation model predicts (i) no intrinsic dependence on growth rate, unless significant back reaction upon slow growth reduces the isotope fractionation towards that characteristic for equilibrium isotope partitioning (this may be observed for Ca isotopes in calcites), (ii) a small decrease of isotope fractionation with increasing temperature that may be amplified if higher temperatures promote back reaction and (iii) a sensitivity to changes in the activation barrier caused by additives such as anions or biomolecules or by the initial formation of amorphous CaCO (sub 3) . Abstract Copyright (2011) Elsevier, B.V.