The oxygen isotope compositions of CO(2) and water vapor samples collected from Larderello geothermal wells after the start of the fluid reinjection program suggest that if the oxygen isotope exchange in the vapor phase does, in fact, exist, it is a very slow process when compared with the residence time of the fluids in the geothermal reservoir. This is because carbon dioxide and water vapor phases could not have equilibrated significantly in the vapor-dominated reservoir. This conclusion implies that the oxygen isotope composition of carbon dioxide may possibly be used as a tool in geothermal exploration for revealing the presence of liquid water in deep geothermal systems. Based on the interpretation of the oxygen isotope data of the CO(2), I propose that the origin of the low oxygen isotope ratios of carbon dioxide at Larderello is the high-temperature exchange with liquid water in the lower reservoir. In Larderello, the liquid water-rock interaction in the lower reservoir may have increased the (18)O/(16)O ratio of the recharge meteoric component. By contrast, lack of high-temperature liquid water in the upper reservoir suggests that the large "delta(18)O shift" described for the upper-reservoir steam during the last decades reflects varying degrees of dilution of the lower-reservoir fluid by the low-(18)O vaporized liquid water of meteoric origin that recharges the field at shallow depth, with local contribution from still deeper high-(18)O water vapor of magmatic origin. The low oxygen isotope composition of the Mesozoic carbonaceous rocks that form the upper reservoir, consequently, likely represents a "fossil" record of the past hot-water geothermal stage. (C) 2011 Elsevier Ltd. All rights reserved.