We present a mechanistic model for the distribution of methane clathrate in marine sediments, and use it to predict the sensitivity of the steady-state methane inventory to changes in the deep ocean. The methane inventory is determined by binning the seafloor area according to water depth, temperature, and O-2 concentration. Organic carbon rain to the seafloor is treated as a simple function of water depth, and carbon burial for each bin is estimated using a sediment diagenesis model called Muds [Glob. Biogeochem. Cycles 16 (2002)]. The predicted concentration of organic carbon is fed into a clathrate model [J. Geophys. Res. 108 (2003)] to calculate steady-state profiles of dissolved, frozen, and gaseous methane. We estimate the amount of methane in ocean sediments by multiplying the sediment column inventories by the corresponding binned seafloor areas. Our estimate of the methane inventory is sensitive to the efficiency of methane production from organic matter and to the rate of fluid flow within the sediment column. Preferred values for these parameters are taken from previous studies of both passive and active margins, yielding a global estimate of 3 x 10 18 g of carbon (3000 Gton C) in clathrate and 2 x 10(18) g (2000 Gton C) in methane bubbles. The predicted methane inventory decreases by 85% in response to 3 degreesC of warming. Conversely, the methane inventory increases by a factor of 2 if the O-2 concentration of the deep ocean decreases by 40 muM or carbon rain increases by 50% (due to an increase in primary production). Changes in sea level have a small effect. We use these sensitivities to assess the past and future state of the methane clathrate reservoir. (C) 2004 Elsevier B.V. All rights reserved.