Recent experiments have shown that radio-frequency (rf) plasma glow discharge using NF3 gas is an effective technique for the removal of uranium oxide from metal surfaces. The results of these experiments are analyzed to explain the measured dependence of the UO2 removal or etch rate on the NF3 gas pressure and the absorbed power in the plasma. The NF3 gas pressure in the experiments was varied from 10.8 to 40 Pa, and the deposited power in the plasma was varied from 25 to 210 W. The UO2 etch rate was strongly dependent on the absorbed power and, to a lesser extent, on the NF3 pressure and decreased exponentially with immersion time. At 210 W and 17 Pa, all detectable UO2 in the samples (~10.6 mg each) was removed at the endpoint, whereas the initial etch rate was ~3.11 m/min. When the absorbed power was 50 W, however, the etch rate was initially ~0.5 g/min and almost zero at the endpoint, with UO2 only partially etched. This self-limiting etching of UO2 at low power is attributed to the formation of nonvolatile intermediates UF2, UF3, UF4, UF5, UO2F, and UO2F2 on the surface. Analysis indicated that the accumulation of UF6 and, to a lesser extent, O2 near the surface partially contributed to the exponential decrease in the UO2 etch rate with immersion time. Unlike fluorination with F2 gas, etching of UO2 using rf glow discharge is possible below 663 K. The average etch rates of the amorphous UO2 in the NF3 experiments are comparable to the peak values reported in other studies for crystalline UO2 using CF4/O2 glow discharge performed at ~150 to 250 K higher sample temperatures.