Recently Hagen et al. (Hagen, W. R., Wassink, H., Eady, R. R., Smith, B. E., and Haaker, H. (1987) Eur. J. Biochem. 169, 457-465) reported the observation of S = 7/2 EPR signals for thionin-oxidized nitrogenase MoFe protein. Here we have studied the protein from Azotobacter vinelandii and Klebsiella pneumoniae with Mössbauer and EPR spectroscopies, with the following results: when the MoFe protein is oxidized by addition of stoichiometric amounts (6-8 equivalents) of dissolved thionin, the well characterized P-cluster state Pox results. Pox has an as yet undetermined, but half-integer electronic spin; however, the state is EPR-silent. In contrast, oxidation by addition of a large excess of solid thionin powder, the method used by Hagen et al., yields mixtures with variable proportions of two oxidized P-cluster forms, namely the familiar Pox and the new state Pox(S = 7/2) observed by Hagen et al. The Mössbauer data suggest that Pox and Pox(S = 7/2) are isoelectronic. The two states, however, have distinct electronic structures; the Mössbauer spectra of Pox exhibit the characteristic trapped-valence Fe2+ site, whereas the spectra of Pox(S = 7/2) lack this feature. Hagen et al. have proposed two new P-cluster models. We conclude that one of the models is incompatible with the Mössbauer data and that the basic assumptions of the other model are not supported by the available data. Finally, the Mössbauer data show that either oxidation method puts the cofactor centers into the diamagnetic state Mox.