Measurements of CO2 and CH4 in 3 m deep groundwater, soil-gas, and soil-atmosphere fluxes were completed in two grasslands in central Italy having the same soil conditions but different subsoil fault-linked secondary permeability. Unfaulted grassland displays gas-phase equilibrium between soil-air and groundwater, typical soil-gas diffusion profiles, and diffusive soil to atmosphere gas transfer; on the basis of oxygen depletion assessment, assuming a ratio of 1:1 between biogenic O2 consumption and CO2 production, the measured soil CO2 concentrations are consistent with a normal production in the soil by biologic activity. The faulted grassland, instead, has higher CO2 and CH4 concentrations (up to 6% and 10 ppmv in soil-air) and flux (1.2 mL m-2 s-1 and 1.3 μL m-2 s-1) resulting from a combination of soil biologic and endogenous components, with evidence of gas transfer from the saturated to the unsaturated zone, and advective gas transfer from soil to the atmosphere. The extra-soil source in the faulted zone, which is about 0.3--4 times the background soil biologic production, is likely related to migrating crustal gas. Whatever the C gas origin and depth may be (biogenic or abiogenic, shallow or deep), the present results demonstrate that the subsoil-derived component occurring in the soil in areas with active tectonics cannot be ignored a priori in the assessment of the C terrestrial sources. In particular, the assumption that dryland soils are sinks for methane, owing to methanotrophic consumption, may be not true in areas affected by active and gas-bearing faults. Accordingly, it would be very important to assess at global scale the actual role in the carbon dioxide and methane cycle of soils within the active tectonic bounds. ¿ 1999 American Geophysical Union