Electron-nuclear double resonance (ENDOR) studies of radical intermediates formed by the oxidative decarboxylation of pyruvate by pyruvate:ferredoxin oxidoreductase were carried out to characterize their electronic structure and elucidate aspects of the recently proposed catalytic mechanism (Menon, S.; Ragsdale, S. W. Biochemistry 1997, 36, 8484-8494). The EPR spectrum of the PFOR/pyruvate adduct at 4 K displays a narrow resonance centered at g = 2.008 that has been attributed to a hydroxyethyl thiamine pyrophosphate (HE-TPP) radical. This spectral feature is superimposed on a broad, complex line shape characteristic of magnetically coupled [Fe4S4] clusters. The ENDOR spectrum at g = 2.008 reveals a broad peak with a complex line shape that can be analyzed, assuming that it arises from a composite of two axially symmetric proton hyperfine couplings. The principle coupling values for these two hyperfine tensors were: A(parallel to)(1) = 18.9 MHz; A(perpendicular to)(1) = 12.6 MHz; and A(parallel to)(2) = 20.3 MHz, A(perpendicular to)(2) = 14.9 MHz. The assignment of these features to the methyl protons of the pyruvate substrate was made using isotopic substitution. The temperature independence of these H-1 ENDOR line shapes from 4 to 200 K indicates that the methyl group of pyruvate undergoes rapid rotation even at 4 K. The ENDOR spectrum at g = 2.008 also shows a pair of derivative peaks centered about the P-31 Larmor frequency that are assigned to a weak hyperfine coupling with the phosphorus nuclei of the TPP cofactor. Two models for the electronic structure of the radical intermediate are discussed. A sigma radical model which postulates a pyruvate-derived acetyl-type radical where little unpaired spin density resides on the TPP cofactor; and a pi radical model that calls for more extensive delocalization of the unpaired electron spin over the HE-TPP framework. Both models require association of the radical center with the pyrophosphate group of TPP to interpret the observed P-31 hyperfine coupling.