Atmospheric carbon dioxide is the raw material for the biosphere. Therefore, changes in the carbon isotopic composition of the atmosphere will influence the terrestrial δ13C signals we interpret. However, reconstructing the atmospheric δ13C value in the geologic past has proven challenging. Land plants sample the isotopic composition of CO2 during photosynthesis. We use a model of carbon isotopic fractionation during C3 photosynthesis, in combination with a meta–data set (519 measurements from 176 species), to show that the δ13C value of atmospheric CO2 can be reconstructed from the isotopic composition of plant tissue. Over a range of pCO2 (198–1300 ppmv), the δ13C value of plant tissue does not vary systematically with atmospheric carbon dioxide concentration. However, environmental factors, such as water stress, can influence the δ13C value of leaf tissue. These factors explained a relatively small portion of variation in the δ13C value of plant tissue in our data set and emerged strongly only when the carbon isotopic composition of the atmosphere was held constant. Members of the Poaceae differed in average δ13C value, but we observed no other differences correlated with plant life form (herbs, trees, shrubs). In contrast, over 90% of the variation the carbon isotopic composition of plant tissue was explained by variation in the δ13C value of the atmosphere under which it was fixed. We use a subset of our data spanning a geologically reasonable range of atmospheric δ13C values (−6.4‰ to −9.6‰) and excluding C3 Poaceae to develop an equation to reconstruct the δ13C value of atmospheric CO2 based on plant values. Reconstructing the δ13C value of atmospheric CO2 in geologic time will facilitate chemostratigraphic correlation in terrestrial sediments, calibrate pCO2 reconstructions based on soil carbonates, and offer a window into the physiology of ancient plants.