Silicon is essential for the growth of diatoms, a group of phytoplankton with opal (amorphous hydrated silica) shells. Diatoms largely control the cycling of silicon in the ocean and, conversely, diatom silica production rates can be limited by the availability of silicic acid. Diatoms are biogeochemically important in that they account for an estimated 75% of the primary production occurring in coastal and nutrient-replete waters1, rising to more than 90% during ice-edge blooms such as occur in the Ross Sea, off Antarctica. There are few means by which to reconstruct the history of diatom productivity and marine silicon cycling, and thus to explore the potential contribution of diatoms to past oceanic biogeochemistry or climate. Indices based on the accumulation of sedimentary opal are often biased by the winnowing and focusing of sediments and by opal dissolution. Normalization of opal accumulation records using particlereactive natural radionuclides may correct for sediment redistribution artefacts and the dissolution of opal within sediments, but not for opal dissolution before it arrives at the sea floor. Half of the opal produced in the euphotic zone may dissolve before sinking to a depth of 200m, constituting a potentially large bias to both normalized and uncorrected records of opal accumulation. Here we exploit the potential that variations in the ratio of 30Si to 28Si in sedimentary opal may provide information on past silicon cycling that is unbiased by opal dissolution. Our silicon stable-isotope measurements suggest that the percentage utilization of silicic acid by diatoms in the Southern Ocean during the last glacial period was strongly diminished relative to the present interglacial.