We have conducted experiments on 16 lunar soils and three lunar pyroclastic glass samples to study the extraction of oxygen, an important resource for a future lunar base. The samples were chosen to span the range of composition, mineralogy, and maturity represented in the Apollo sample collection. Each sample was reduced in flowing hydrogen for 3 hours at 1050°C. The dominant effect was reduction of Fe2+ (as FeO) in minerals and glass to iron metal, with concomitant release of oxygen. Oxygen extraction was strongly correlated with initial Fe2+ abundance. The data and supporting experiments also indicate extraction of oxygen from TiO2 and SiO2. No dependence of oxygen release on sample maturity could be identified. Oxygen was extracted, in decreasing order of efficiency, from ilmenite, agglutinitic and pyroclastic glass, olivine, and pyroxene. The experimental reduction of lunar soil provides a method for assessing the oxygen production potential for sites on the lunar surface from lunar orbit. Our results show that oxygen yield from the regolith can be predicted from knowledge of only one parameter, total iron content. This parameter can be measured from orbit by gamma ray spectrometry or multispectral imaging. These experimental data also support the selection of feedstocks and operating parameters for a lunar oxygen plant.