The silicon isotope composition of biogenic silica (delta Si-30(BSi)) in the ocean is a function of the delta Si-30 of the available dissolved Si (DSi; H2SiO4), the degree of utilisation of the available DSi, and, for some organisms, the concentration of DSi. This makes delta Si-30(BSi) in sediment archives a promising proxy for past DSi concentrations and utilisation. At steady-state, mean delta Si-30(BSi) must equal a weighted average of the inputs, the majority of which are of continental origin. Variation in the functioning of the continental Si cycle on timescales similar to the residence time of DSi in the ocean (similar to 10 ka) may therefore contribute to downcore variability in delta Si-30(BSi) on millennial or longer time-scales. The direction and magnitude of change in published delta Si-30(BSi) records over the last few glacial cycles is consistent among ocean basins and between groups of silicifiers. They document glacial values that are typically 0.5 to 1.0 parts per thousand lower than interglacial values and together hint at coherent and predictable glacial-interglacial variability in whole-ocean delta Si-30 driven by a change in mean delta Si-30 of the inputs. In this contribution, we review the modern inputs of DSi to the ocean and the controls on their isotopic composition, and assess the evidence for their variability on millennial-plus timescales. Today, 9.55 x 10(12) mol yr(-1) DSi enters the ocean, of which roughly 64% and 25% are direct riverine inputs of DSi, and DSi from dissolution of aeolian and riverborne sediment, respectively. The remainder derives from alteration or weathering of the ocean crust. Each input has a characteristic delta Si-30, with our current best estimate for a weighted mean being 0.74 parts per thousand, although much work remains to be done to characterise the individual fluxes. Many aspects of the continental Si cycle may have differed during glacial periods that together can cumulatively substantially lower the mean delta Si-30 of DSi entering the ocean. These changes relate to i) a cooler, drier glacial climate, ii) lowered sea level and the exposure of continental shelves, iii) the presence of large continental ice-sheets, and iv) altered vegetation zonation. Using a simple box-model with a Monte-Carlo approach to parameterisation, we find that a transition from a hypothesised glacial continental Si cycle to the modern Si cycle can drive an increase in whole ocean delta Si-30 of comparable rate and magnitude to that recorded in delta Si-30(BSi). This implies that we may need to revisit our understanding of aspects of the Si cycle in the glacial ocean. Although we focus on the transition from the last glacial, our synthesis suggests that the continental Si cycle should be seen as a potential contributory factor to any variability observed in ocean delta Si-30(BSi) on millennial or longer timescales.