This study reports the first field test of a multi-channel, auto-dilution, steady-state, soil–CO2 flux monitoring system being developed to help understand the pathways by which fugitive CO2 from a geologic sequestration site migrates to the surface. The test was conducted from late August through mid-October 2008 at the Zero Emissions Research and Technology project site located in Bozeman, MT. Twenty steady-state and five non-steady-state flux chambers were installed in a 10 × 15 m area, one boundary of which was directly above a shallow (2-m depth) horizontal injection well located 0.5 m below the water table. A total flux of 52 kg CO2 day−1 was injected into the well for 13 days and the efflux from the soil was monitored by the chambers before, during, and for 33 days after the injection. The results showed a rapid increase in soil efflux once injection started, with maximal values reached within 3–7 days in most chambers. Efflux returned to background levels within a similar time period after injection ceased. A radial efflux pattern was observed to at least 2 m from the injection well, and evidence for movement of the CO2 plume during the injection, presumably due to groundwater flow, was seen. The steady-state chambers yielded very stable data, but threefold to fivefold higher fluxes than the non-steady-state chambers. The higher fluxes were attributed to vacuum induced in the steady-state chambers by narrow vent tubes. High winds resulted in significant decreases in measured soil CO2 efflux, presumably by enhancing efflux from soil outside the chambers. [ABSTRACT FROM AUTHOR]