Extensive studies of the physiological protein—protein electron-transfer (ET) complex between yeast cytochrome c peroxidase (CcP) and cytochrome c (Cc) have left unresolved questions about how formation/dissociation of binary and ternary complexes influence ET. We probe this issue through study of the photocycle of ET between Zn-ProtoporphyrinIX-substituted CcP(W191F) (ZnPCcP) and Cc. Photoexcitation of ZnPCcP in complex with Fe3+Cc, initiates the photocycle: charge-separation ET [3ZnPCcP, Fe3+Cc]→[ZnP+CcP, Fe2+Cc] followed by charge recombination, [ZnP+CcP, Fe2+Cc] → [ZnPCcP, Fe3+Cc]. The W191F mutation eliminates fast hole hopping through W191, enhancing accumulation of charge-separated intermediate and extending the timescale for binding/dissociation of the charge-separated complex. Both triplet quenching and the charge-separated intermediate were monitored during titrations of ZnPCcP with Fe3+Cc, Fe2+Cc, and redox-inert CuCc. The results require a photocycle that includes dissociation/recombination of the charge-separated binary complex and a charge-separated ternary complex, [ZnP+CcP, Fe2+Cc, Fe3+Cc]. The expanded kinetic scheme formalizes earlier proposals of “substrate-assisted product dissociation” within the photocycle. The measurements yield the thermodynamic affinity constants for binding the first and second Cc: KI = 10−7 M−1, KII = 10−4 M−1. However, two-site analysis of the thermodynamics of formation of the ternary reveals that Cc binds at the weaker-binding site with much greater affinity than previously recognized, and places upper bounds on the contributions of repulsion between the two Cc of the ternary complex. In conjunction with recent NMR studies, the analysis further suggests a dynamic view of the ternary complex, wherein neither Cc necessarily faithfully adopts the crystal-structure configuration because of Cc-Cc repulsion.