This is a summarizing and reviewing report in which almost all the experimental data representing Project work done prior to 1949 on the corrosion of pure uranium are brought together and analyzed. New data obtained in this laboratory on corrosion rates in laboratory atmosphere and on the identification of corrosion products by electron diffraction are included. The data for corrosion in each of several different media have been plotted according to logarithm-of-the-rate versus reciprocal-temperature coordinates; and from these plots values for the energies and entropies of activation of the corrosion reactions have been obtained. By theoretical treatment of the role of oxygen as a negative catalyst, it is shown that this element may be expected to "poison" the corrosion rection and then act as a corrosion inhibitor. A practical significance of these data analyses is that they explain why machined uranium parts stored in inert atmospheres (helium or argon) containing only very small amounts of water vapor will corrode relatively rapidly with the formation of loose powdery oxide, whereas similar parts may be kept in dry air almost indefinitely with formation on their surface of only a discoloring, but adhered, oxide coating.