Abstract The political and social acceptability of the geological storage of CO 2 depends on the existence of reliable site surveillance technologies. The ground surface is the widest available area over which a permanent control of possible leaks may be performed, and also a privileged area where to investigate to which extent a leak at the reservoir depth may disturb the superficial environment. In this article, we suggest several surface surveillance guidelines inspired from data originating from an EOR- CO 2 oil field case study where an extensive survey of surface and reservoir gas has been performed. This study shows that the gases that may leak from a CO 2 geological storage can have a wide range of compositions, both molecular and isotopic. The complexity in the gas source is further impacted by the likely changes in gas (or more generally fluid) composition on the leaking pathway. With such heterogeneity and variability in mind, the task of monitoring a leak at the surface becomes a serious challenge. Furthermore, comparing natural background fluxes of CO 2 with expected fluxes in worst case leak scenarios shows that both have comparable amplitudes (∼0.4 mol/m 2 /day). The flux of CO 2 itself, or the CO 2 content of the soil are of limited interest with respect to a leak detection. Using calculations based on the Dusty Gas Model, we show that the coupling of carbon isotopes and the noble gases with flux measurements is a powerful monitoring method, able to discriminate the flux contribution of a leak in the flux range of the natural baseline.