The hydrogenase (EC 1.2.2.1) of Desulfovibrio gigas is a complex enzyme containing one nickel center, one [3Fe-4S] and two [4Fe-4S] clusters. Redox intermediates of this enzyme were generated under hydrogen (the natural substrate) using a redox-titration technique and were studied by EPR and Mössbauer spectroscopy. In the oxidized states, the two [4Fe-4S]2+ clusters exhibit a broad quadrupole doublet with parameters (apparent delta EQ = 1.10 mm/s and delta = 0.35 mm/s) typical for this type of cluster. Upon reduction, the two [4Fe-4S]1+ clusters are spectroscopically distinguishable, allowing the determination of their midpoint redox potentials. The cluster with higher midpoint potential (-290 +/- 20 mV) was labeled Fe-S center I and the other with lower potential (-340 +/- 20 mV), Fe-S center II. Both reduced clusters show atypical magnetic hyperfine coupling constants, suggesting structural differences from the clusters of bacterial ferredoxins. Also, an unusually broad EPR signal, labeled Fe-S signal B', extending from approximately 150 to approximately 450 mT was observed concomitantly with the reduction of the [4Fe-4S] clusters. The following two EPR signals observed at the weak-field region were tentatively attributed to the reduced [3Fe-4S] cluster: (i) a signal with crossover point at g approximately 12, labeled the g = 12 signal, and (ii) a broad signal at the very weak-field region (approximately 3 mT), labeled the Fe-S signal B. The midpoint redox potential associated with the appearance of the g = 12 signal was determined to be -70 +/- 10 mV. At potentials below -250 mV, the g = 12 signal began to decrease in intensity, and simultaneously, the Fe-S signal B appeared. The transformation of the g = 12 signal into the Fe-S signal B was found to parallel the reduction of the two [4Fe-4S] clusters indicating that the [3Fe-4S]o cluster is sensitive to the redox state of the [4Fe-4S] clusters. Detailed redox profiles for the previously reported Ni-signal C and the g = 2.21 signal were obtained in this study, and evidence was found to indicate that these two signals represent two different oxidation states of the enzyme. Finally, the mechanistic implications of our results are discussed.