Gram-scale experiments on the carbothermic reduction of UO/sub 2/ were carried out as part of an investigation of oxidation-reduction reprocessing of UC. UO/sub 2/, from air oxidation of UC, was mixed with graphite powder and heated in vacuum at 1300 to 1700 deg C, both as loose powder and as pellets. The best container material was carbide-coated graphite; molybdenum was satisfactory with pellets. Information on reaction rates was obtained by measuring the evolved carbon monoxide with a gas meter. In the powder experiments, the carbon content of the products agreed with predicted values, with a standard deviation of 0.069 wt%. The carbon content of products from pellets was low by 0.25 wt%, with a standard deviation of 0.076 wt%. This is believed to be an effect of the acrylic plastic binder. Reaction rates followed a second-order rate equation, except that thc rate constant often decreased abruptly at about 90% completion. Correlation between rate and temperature was poor; for 32 values of the rate constant, K (reciprocal hours), the least squares equation was: log K = 9.8216,080/T, with a standard error of 0.20 in log K. The data also agree fairly well (standard error, 0.22) with the hypothesis that the activation energy is equal to the heat of the endothermic reaction. Because of variations in the rate, it is recommended that the reaction of each batch should be followed with a gas meter, in order to obtain a desired degree of completion. (auth).