The presence of sodium and potassium on the Moon implies that other more abundant species should be present. Volatile molecules like H2O are significantly more abundant than sodium in any of the proposed external atmospheric sources. Source mechanism which derive atoms from the surface should favor abundant elements in the regolith. It is therefore puzzling that the Apollo ultraviolet spectrometer experiment set limits on the density of oxygen of NO<5¿102 cm-3, and that the Apollo Lunar Atmospheric Composition Experiment data imply NO<50 cm-3 above the subsolar point. These limits are surprisingly small relative to the measured value for sodium. A simple consideration of sources and sinks predicts significantly greater densities of oxygen. It is possible but doubtful that the Apollo measurements occurred during an epoch in which source rates were small. A preferential loss process for oxygen on the darkside of the Moon is considered in which ionization by electron capture in surface collisions leads to escape through acceleration in the local electric field. Cold trapping in permanently shadowed regions as a net sink is considered and discounted, but the episodic nature of cometary insertion may allow formation of ice layers which act as a stabilized source of OH. On the basis of an assumed meteoroid impact source we predict a possible emission brightness of ~50 R in the (OH(A-X)(0,0) band above the lunar bright limb. A very uncertain small comet source of H2O could raise this value by more than two orders of magnitude. ¿ American Geophysical Union 1991