Identification of the source of CO 2 in natural reservoirs and development of physical models to account for the migration and interaction of this CO 2 with the groundwater is essential for developing a quantitative understanding of the long term storage potential of CO 2 in the subsurface. We present the results of 57 noble gas determinations in CO 2 rich fields (>82%) from three natural reservoirs to the east of the Colorado Plateau uplift province, USA (Bravo Dome, NM., Sheep Mountain, CO. and McCallum Dome, CO.), and from two reservoirs from within the uplift area (St. John's Dome, AZ., and McElmo Dome, CO.). We demonstrate that all fields have CO 2 / 3 He ratios consistent with a dominantly magmatic source. The most recent volcanics in the province date from 8 to 10 ka and are associated with the Bravo Dome field. The oldest magmatic activity dates from 42 to 70 Ma and is associated with the McElmo Dome field, located in the tectonically stable centre of the Colorado Plateau: CO 2 can be stored within the subsurface on a millennia timescale.The manner and extent of contact of the CO 2 phase with the groundwater system is a critical parameter in using these systems as natural analogues for geological storage of anthropogenic CO 2 . We show that coherent fractionation of groundwater 20 Ne/ 36 Ar with crustal radiogenic noble gases ( 4 He, 21 Ne, 40 Ar) is explained by a two stage re-dissolution model: Stage 1: Magmatic CO 2 injection into the groundwater system strips dissolved air-derived noble gases (ASW) and accumulated crustal/radiogenic noble gas by CO 2 /water phase partitioning. The CO 2 containing the groundwater stripped gases provides the first reservoir fluid charge. Subsequent charges of CO 2 provide no more ASW or crustal noble gases, and serve only to dilute the original ASW and crustal noble gas rich CO 2 . Reservoir scale preservation of concentration gradients in ASW-derived noble gases thus provide CO 2 filling direction. This is seen in the Bravo Dome and St. John's Dome fields. Stage 2: The noble gases re-dissolve into any available gas stripped groundwater. This is modeled as a Rayleigh distillation process and enables us to quantify for each sample: (1) the volume of groundwater originally 'stripped' on reservoir filling; and (2) the volume of groundwater involved in subsequent interaction. The original water volume that is gas stripped varies from as low as 0.0005 cm 3 groundwater/cm 3 gas (STP) in one Bravo Dome sample, to 2.56 cm 3 groundwater/cm 3 gas (STP) in a St. John's Dome sample. Subsequent gas/groundwater equilibration varies within all fields, each showing a similar range, from zero to $100 cm 3 water/cm 3 gas (at reservoir pressure and temperature).