In the case of producing large amounts of hydrogenous gas, currently there are no problems related to basic techniques of hydrogen production and distribution, but the main technological problem will consists of storing it in order to regulate the difference between permanent or increasing gas production and seasonally modulated gas consumption. The most efficient and most inexpensive method of storing large amounts of hydrogen is to inject them in geological formations like aquifers, depleted gas reservoirs, or salt caverns (Zittel and Wurster 1996). The cost is of order $ 3.5 per 1 GJ (Taylor et al. 1986). Several underground storages of hydrogen (USH) or town gas exist in the word, as for instance, Teeside in the UK, in Texas, in Russia, Kiel in Germany, Lobodice in Czechoslovakia, Beynes - an ex-storage in France. During several tens of years the storage of hydrogen was considered as something deja-vu, to be similar to that of natural gas, which is amplified by the chemical inactivity of hydrogen and its very low solvability in groundwater [Bulatov 1979; Carden and Paterson 1979; Lindblom 1985; Paterson 1983]. Nevertheless, quite unusual behaviour of UHS was discovered by in situ monitoring of the gas composition extracted during the cycle "production" which followed the cycle "injection". These observations (Smigai et al. 1990; Buzek et al. 1994) revealed high variations of gas composition in time and space. In particular, a significant reduction in the H2 and CO2 contents and a simultaneous increase in CH4 contents were observed in the Lobodice town gas storage facility (Smigai et al. 1990). Similar phenomena were recorded in Beynes. After several months of injection and storage, at the beginning of the cycle "production" the twofold increase of the methane contents in the reservoir gas and the twofold reduction of the CO2 CO contents was observed. The contents of hydrogen decreased by 1.4. The explanation to these observations has been done in (Buzek et al. 1994) in terms of the in situ methane generators by methanogenic bacteria which catalyse the reaction between hydrogen and CO2/CO, by producing methane and water. Further observations have revealed even more unusual effects within the storage facility, such as creating a spatial alternation of the areas saturated preferably by hydrogen or methane. This proved an in situ natural separation of chemical components in space. Thus, we are dealing with a natural reactor which partially destroys CO2 and H2 and doubles the mass of methane. It is clear that the problem is important for industry as it concerns both the energy sector and ecology. The resulting economical efficiency of such a process can be estimated only after the physical and mathematical modeling of all possible scenarios of the reservoir behaviour. The development of such a model represents the main objective of this paper.