We propose to use ISO observations of the 63 micron line of OI to
 determine the gas phase oxygen abundance in dense interstellar
 clouds, where complex molecules are found and where the interstellar
 UV radiation field is attenuated by dust. It has been shown that
 carbon can be largely accounted for by CO and CI measurements,
 but the predicted molecular reservoirs of oxygen, O_{2} and H_{2}O,
 have been shown not to account for the anticipated gas phase oxygen
 by at least an order of magnitude. We expect that the remainder is
 largely atomic oxygen and propose a method of 63 micron absorption
 spectroscopy to identify the missing oxygen content of these clouds.
 In order to separate OI emission from the background continuum source
 from OI absorption by the foreground source we select lines of sight
 where the velocity separation of emitter and absorber is greater
 than the LWS resolution of 30 km/s. Also, to avoid the effects of
 strong PDRs we choose absorbing clouds well removed from ionizing
 stars. Further, to aid in the understanding of the relation between
 the oxygen generated in the PDRs at the cloud surfaces, caused by
 the ambient interstellar UV field, and oxygen associated with the
 more shielded molecular region, we measure a set of clouds with a
 range of H_{2} column densities. The cloud OI absorption should
 correlate with the H_{2} column densities (determined from trace
 molecules), whereas the PDR contribution should be roughly a
 constant.