Abstract A vital aspect to public and regulatory acceptance of carbon sequestration is assurance that groundwater resources will be protected. Theoretical and laboratory studies can, to some extent, be used to predict the consequences of leakage. However, direct observations of CO 2 flowing through shallow drinking water aquifers are invaluable for informing credible risk assessments. To this end, we have sampled shallow wells in a natural analog site in New Mexico, USA, where CO 2 from natural sources is upwelling from depth. We collected major ion, trace element, and isotopic ( 3 H, 18 O, and Sr) data and, coupled with laboratory experiments and reactive transport modeling, have concluded that the major control on groundwater quality at this site is not chemical reaction of CO 2 with the aquifer but intrusion of saline waters upwelling with the CO 2 . Using reactive transport modeling based on field data, we show the difference in reactivity of the CO 2 and CO 2 /saline water source terms, particularly with respect to carbonate mineralogy. Sr isotopes were used to investigate whether aquifer waters were affected by carbonate mineral reaction with CO 2 or by saline water intrusion. Preliminary data suggest that Sr isotopes can successfully be used to discriminate between the two types of source terms at Chimayo; this technique shows promise for monitoring CCS sites. In developing predictive capabilities for future sites, it is critical to identify the solid phases and specific reactions controlling dissolved trace metal concentrations in both the presence and absence of CO 2 . We have conducted laboratory experiments to identify these phases and have found that some elements (e.g., U, Ca) are largely controlled by ion exchange and/or carbonate minerals. In the experiments, the concentration of some metals increases after exposure to CO 2 (although concentrations remain below the U.S. EPA primary drinking water standards); we are currently extending these experiments to determine if the reactions causing the increase are reversible and, if so, on what time scales. Metal scavenging by secondary mineral precipitation, as observed at other natural analog sites, may be important at certain temporal scales. We are using the information gained from this field and laboratory study to develop predictive models for application to risk assessment at future CCS sites. The models will be particularly useful in identifying the temporal and spatial scales of water quality changes and in developing possible mitigation strategies in the case of leaks at engineered CCS sites.