Using broken-symmetry density functional theory, we have studied an experimentally proposed model for ribonucleotide reductase (RNR) intermediate X, which contains a single oxo bridge, one terminal H(2)O or OH(-) ligand, a bidentate carboxylate from Glu115, and a mono-oxygen bridge provided by Glu238. For the models proposed here, the terminal H(2)O/OH(-) ligand binds to site Fe1 which is closer to Tyr122. The diiron centers are assigned as high-spin Fe(III)Fe(IV) and antiferromagnetically coupled to give the S(total) = (1)/(2) ground state. Calculations show that the model with a terminal hydroxide in the antiferromagnetic [S(Fe1) = 2, S(Fe2) = (5)/(2)] state (Fe1 = Fe(IV), Fe2 = Fe(III)) is the lowest energy state, and the calculated isomer shift and quadrupole splitting values for this cluster are also the best among the four clusters studied here when compared with the experimental values. However, the DFT-calculated (1)H proton and (17)O hyperfine tensors for this state do not show good agreement with the experiments. The calculated Fe1-Fe2 distances for this and the other three clusters at >2.9 A are much longer than the 2.5 A which was predicted by the EXAFS measurements. The mono-oxygen bridge provided by Glu238 tends to be closer to one of the Fe sites in all clusters studied here, and it does not function as a bridge in helping to produce a short Fe-Fe distance. Overall, the models tested here are not likely to represent the core structure of RNR intermediate X. The model with the terminal OH(-) binding to the Fe1(III) center shows the best calculated (1)H proton and (17)O hyperfine tensors compared with the experimental values. This supports the earlier proposal based on analysis of ENDOR spectra (Willems et al.(16)) that the terminal oxygen group binds to the Fe(III) site in RNR-X.