We propose a model for geminate electron–hole dissociation in organic photovoltaic (OPV) cells and show how power conversion efficiencies greater than those currently achieved might be realized via design strategies employing moderate optical bandgaps and enhanced charge delocalization near the donor–acceptor interface. Applying this model to describing geminate electron–hole dissociation via charge transfer (CT) states, we find good agreement with recently published high-efficiency experimental data. The optimal bandgap for current-generation organic active layer materials is argued to be 1.7 eV – significantly greater than in previous analyses, including the Shockley–Queisser approach based upon non-excitonic solar cell dynamics. For future higher efficiency OPVs, the present results show that the optimal bandgap should be slightly lower, 1.6 eV. Finally, these results support design strategies aimed at enhancing mobility near the donor–acceptor interface and reducing the electron–hole binding energy, rather than striving to further reduce the bandgap.