Atmospheric CO2 variations on orbital timescales mainly depend on variations in the carbon content of the ocean (including its sediments) and of the terrestrial biosphere. However, the input fluxes to this subsystem of the global carbon cycle (namely volcanic CO2 emissions, oxidative weathering of organic matter and carbonate weathering), and the outgoing fluxes (burial of organic matter and carbonate sediments) result in a carbon turnover of the ocean-atmosphere-biosphere system of only 100 kyr. Without negative feedbacks, even small flux imbalances would lead to a gradual rise or drop in mean atmospheric CO2 over the last 800 kyrs that is not observed (Bereiter et al., 2015). Weathering is thought to provide this negative feedback since weathering is sensitive to temperature and runoff. Here we present a new, ice core-based record of tetrafluoromethane (CF4) that serves as proxy for silicate weathering, since it is released only from weathering of granites. CF4 is a very inert gas with an atmospheric lifetime on the order of 100 kyr and therefore it accumulates in the atmosphere and can be measured in air bubbles in ice cores. Our CF4 concentration record from the EPICA Dome C ice core covers the last 800 kyr and reveals clear glacial/interglacial cyclicity. Owing to its long lifetime, CF4 concentrations have to be inverted into CF4 fluxes to yield a metric proportional to changes in granite weathering. This CF4 flux shows a tight positive correlation with CO2 and temperature suggesting the existence of a glacial/interglacial response in global granite weathering to climate change. Furthermore, we identify an intensification in weathering after the Mid Brunhes Event (MBE) ca. 430 kyr ago. On average, global CF4 emissions after the MBE, and thus the inferred weathering rate, are higher by about 15% than before the MBE. As the MBE affects only the intensity of interglacials, which typically comprise less than 30% of the entire glacial/interglacial cycle, the 15% rise in weathering after the MBE is connected to these short interglacials only. Accordingly, our data suggest that the global weathering rate may be rather sensitive to climate- related changes, either directly through temperature or runoff for a given area, through enhanced chemical weathering following glacial erosion, or indirectly by increasing the size of the affected land area. Consequently, any long-term imbalance in carbon or alkalinity fluxes which would otherwise cause a drift in mean CO2 is counter-balanced by the sensitivity of weathering rates. Our data thus support the observations of a declining atmospheric O2 level over the last 800 kyr (Stolper et al., 2016) while at the same time CO2 during the Quaternary stayed within narrow boundaries.