At an average location on the surface of Mars, the pressure of CO 2 ( P CO 2 ) varies seasonally between about 6 and 8 mbar. Outgassing models suggest that at least 140 mbar, and possibly as much as 3000 mbar, of CO 2 have been placed in the atmosphere over geologic time. Neither the polar caps nor the regolith alone appears to be an adequate repository for the CO 2. The north polar cap does not contain a permanent store of CO 2 ice, while the south polar cap is unlikely to hold in excess of a few millibars. The regolith can adsorb no more, and probably much less, than 280 mbar of CO 2. Mechanisms associated with storing carbon in these reservoirs do not account for the particular range and stability of P CO 2 found on Mars today. It is argued that carbonate rock is the most reasonable reservoir for the excess CO 2 and that the rock formation process can explain the current CO 2 pressure. To effect carbonate formation at a rate rapid enough to produce significant deposits over geologic time, liquid water, at least in transitory pockets, is apparently required. Solid-solid and solid-gas reactions are probably orders of magnitude too slow; however, even intermittent aqueous chemistry may be sufficient. Cations are also needed, and existing constraints on the chemical state of the Martian soil do not preclude their occurrence in usable forms and adequate supply. A feedback mechanism that links the evolution of P CO 2 directly to the occurrence of liquid water is postulated. According to the scenario, the evolution of P CO 2 is controlled largely by aqueous chemistry forming carbon-containing sedimentary rocks as on Earth, perhaps during early history in open water, but more recently in transitory pockets of moisture in the soil. Once the total atmospheric pressure is reduced to near a limiting value ( P ★), below which liquid water can not form in the Mars environment, the occurrence of transitory pockets is inhibited, and atmospheric CO 2 is no longer depleted by an efficient mechanism. While present conditions on Mars preclude the existence of open bodies of liquid water, the formation of moisture in disequilibrium is not excluded by any known constraints. To a first approximation, the water evaporation rate is inversely proportional to P CO 2 , which suggests the existence of a minimum overburden pressure P ★. Calculations showing that it is difficult but not impossible to form liquid water in disequilibrium on Mars today support the argument that a limiting value of P CO 2 has been approached. P CO 2 is currently quite close to the triple-point pressure of water, which is the minimum equilibrium vapor pressure of water above the pure liquid. This too is in accord with the hypothesis, given existing constraints on current Martian conditions, since the minimum disequilibrium overburden pressure is unlikely to be lower, but need not be much higher, than the triple-point pressure of water if a feedback mechanism of the type proposed is operating. To form transitory pockets of pure liquid water, the smallest possible value of P ★ is loosely constrained to 6.1 mbar by the minimum criterion, in a space- and time-averaged sense, for which very rapid evaporation of ice would occur until open liquid water could be maintained in equilibrium on the surface. From a consideration of available insolation at Mars, a very crude upper bound on P ★ of around 30 mbar is obtained. The hypothesis has profound implications for the history of the Mars surface and atmosphere. It suggests that unless rapid outgassing events occurred subsequent to early Mars times, the climate of the planet has, to first order, evolved linearly, rather than in the cyclic manner allowed by polar cap storage or some models of regolith adsorption reservoirs. Substantial carbonate deposits are predicted, though they may be mixed in the soil or covered by eolian debris. If these deposits can be mapped on a global scale, their spatial distribution should contain clues about the way heat and water needed for sedimentation are supplied.