When CO2 is injected into storage reservoirs, buoyancy drives flow upward towards overlying caprock layers that retard flow. Caprocks are commonly shale formations with extremely low pore connectivity, but they contain a large amount of nanopores (pore width < 200 nm). Nanopores with large surface area can potentially adsorb and store any CO2 molecules that manage to flow into the caprock. The heterogenous shale composition, with varying chemical and physical structures, controls both wetting and sorption properties. The goal of this study is to determine the extent to which mineral attributes affect CO2 accessible pore spaces, using low-pressure adsorption. We use pure clay samples and 55 shale samples that are from Bakken, Wolfcamp, Utica, Niobrara, and Green River formations in the US, and from Agardhfjellet and Rurikfjellet formations in Svalbard. We show that organic matter and clay groups are the primary drivers of CO2 adsorption. Organic matter has a large amount of micropores for storage and a natural affinity for CO2, while clay minerals store CO2 on the large specific surface area provided by the expandable clays (smectite-illite-mica), mostly in meso- to macropores. CO2 adsorption increases with thermal maturation of both organic matter and clay minerals. Inorganic minerals, such as carbonate, quartz, K-feldspar, plagioclase and pyrite, are non-sorbing minerals that have an inverse relation to CO2 adsorption. However, chemisorbed water in calcite mineral interlayers can dissolve CO2. Due to competing and complementing effects of mineral composition, correlations of CO2 storage capacity and lithology is best understood using multivariate analysis.