Monitoring the surface composition of CO2 derived from subsurface reservoirs is an important part of the carbon capture and storage (CCS) chain. Most approaches use geochemical or geophysical instrumental approaches but these have the drawback that no long-term time series are available, which depend on a predefined monitoring location. We test a flexible approach using natural archives based on the measurement of radiogenic 14C concentrations in tree rings to detect geogenic CO2 fluxes derived from natural springs in the Latera Caldera, central Italy. The approach can be used as a preliminary check to evaluate natural CO2 leaks from sites designated for CO2 storage, as well as evaluating the extent of leakage in an unforeseen area. An extensive database of soil gas composition and fluxes is available for that site, permitting direct comparison of the tree ring isotopic composition and point sources of CO2 from the subsurface against the mean atmospheric standard. We sampled oak trees (Quercus cerris and Quercus pubescens) directly at the CO2 source (ON), and at short (50m, NEAR), intermediate (500m, FAR) and long distances (~3000m, CONTROL) from CO2, sources, and measured the radioactive 14C concentration in tree rings at ~10 year intervals from 2012 back to 1976. To accurately date the tree rings we constructed a tree-ring chronology using standard dendrochronological methods. We tested whether variation in 14C concentration in tree rings and ring-width are related to distance of trees from CO2 source, as well as climate factors, i.e. precipitation and temperature. Results show that local point sources of CO2 at the location where the tree grows are effectively recorded by the 14C concentration in the cellulose of this tree. The fossil CO2 signal is sharply delineated since already at short distance from the source (~50m, NEAR) the 14C incorporation is at the detection limit of the tested approach. Tree-ring width of the oaks at Caldera Latera is mainly limited by the amount of precipitation during the growing season, from March to October, while distance to the CO2 point source has no detectable effect on radial growth, likely due to the continuous presence of the enhanced CO2 flux to which the trees adapt their physiology from germination. While the approach is promising and permits data collection at any forested site, a more detailed sampling transect between 0 and 50m from a point source of CO2 is needed to determine the exact detection limit of the signal in the 14C concentration of the tree ring cellulose. © 2013 Elsevier Ltd.