Although the cytochrome c peroxidase/H2O2 reaction product, compound ES, has been a long-standing subject of research, only recently has its broad EPR signal been proven to arise from a radical at Trp-191. Despite this advance, no model has satisfactorily explained the anomalous breadth and shape of this signal, which is conventionally interpreted as having axial symmetry with g parallel approximately 2.04 > g perpendicular approximately 2.01, contrary to expectations for a planar pi radical. Furthermore, these g values exhibit marked temperature and preparation dependencies as well as an unexplained high-field "tail" extending from the g = 2.01 peak. We have reexamined the EPR and ENDOR spectra of compound ES at 35 GHz, as well as those of compound ES in the mutant D235E. This mutation significantly alters the line shape of the Trp-191 free radical. We present a comprehensive model that successfully accounts for the properties of this unusual protein free radical. We show that the EPR spectra of both proteins can be described in terms of a weak exchange interaction between the S = 1 oxyferryl (Fe = O)2+ moiety and a radical on Trp-191; a distribution in protein conformation leads to a distribution in the coupling, which ranges from ferromagnetic to antiferromagnetic. We also derive, for the first time, explicit expressions for frozen-solution and single-crystal spectra of such spin-coupled systems and show that the model accounts for all the data that previously led to apparent anomalies in the interpretation of the frozen-solution and single-crystal [Hori, H., & Yonetani, T. (1985) J. Biol. Chem. 260, 349-355] EPR properties. Finally, we have used the CW EPR and pulsed-EPR saturation-recovery methodology to address reports that the broad signal from the spin-coupled Trp-191 radical is accompanied by a minority (approximately 10%), narrow signal that is associated with a radical site other than Trp-191. We find no evidence for such a species and discuss the earlier reports in light of our model.