A thermodynamic model for hydrate formation is used to compute the solubility of methane in pore water in equilibrium with gaseous methane or methane hydrate or both. Free energy of water in the hydrate phase and of methane in gas bubbles are corrected to account for salt effects and capillary effects. Capillary effects increase the solubility of methane in fluid in equilibrium with either hydrate or gas. Natural sediments have a broad distribution of pore sizes, and the effective pore size for capillary effects is a function of the fraction of the pore space filled by hydrate or gas (phase fraction). The equilibrium conditions for hydrate+water+gas equilibrium thus depend on hydrate and gas phase fraction. Data acquired on Blake Ridge during Ocean Drilling Program Leg 164 show that the base of the hydrate stability there is shifted by -2 ¿C or more with respect to the expected temperature and this shift has been attributed to capillary effects. We show that this explanation would require a very small effective pore radius (20 nm at 30 MPa). Mercury porosimetry indicates that the percolation threshold for Blake Ridge silty claystone is reached at 20--25% phase fraction and corresponds to a 100 nm pore radius. Hydrate and gas phase fraction determined with several independent methods are all lower than this percolation threshold, implying that gas and hydrate fill pores larger than 100 nm. We conclude that additional inhibition factors other than pore size effects must be involved to explain the -2 ¿C bottom-simulating reflector (BSR) shift as an equilibrium phenomenon. Capillary effects may, however, explain other observations such as large variations of the gas hydrate content in the sediment with lithology and porosity and the distribution of hydrate between interstitial hydrate and segregated masses. Capillary effects should also oppose the migration of gas bubbles when gas phase fraction is less than the percolation threshold and make unnecessary the assumption of a hydrate seal impermeable to fluids. Alternatively, we can go some way to explaining the offset position of the BSR by relaxing the assumption that the system is in thermodynamic equilibrium. Nucleation kinetics of hydrate and/or free gas bubbles may be inhibited by confinement of the methane-bearing fluid in small pores. Equilibration may also be limited by possible rates of diffusional transport of gas, water, and salt components or be perturbed by significant flows of fluid or heat through the sediments. ¿ 1999 American Geophysical Union