We report herein a comparison of the photophysics of a series of polythiophenes with ionization potentials ranging from 4.8 to 5.6 eV as pristine films and when blended with 5 wt % 1-(3-methoxycarbonyl)propyl-1-phenyl-[6,6]C61 (PCBM). Three polymers are observed to give amorphous films, attributed to a nonplanar geometry of their backbone while the other five polymers, including poly(3-hexylthiophene), give more crystalline films. Optical excitation of the pristine films of the amorphous polymers is observed by transient absorption spectroscopy to give rise to polymer triplet formation. For the more crystalline pristine polymers, no triplet formation is observed, but rather a short-lived (approximately 100 ns), broad photoinduced absorption feature assigned to polymer polarons. For all polymers, the addition of 5 wt % PCBM resulted in 70-90% quenching of polymer photoluminescence (PL), indicative of efficient quenching of polythiophene excitons. Remarkably, despite this efficient exciton quenching, the yield of dissociated polymer+ and PCBM- polarons, assayed by the appearance of a long-lived, power-law decay phase assigned to bimolecular recombination of these polarons, was observed to vary by over 2 orders of magnitude depending upon the polymer employed. In addition to this power-law decay phase, the blend films exhibited short-lived decays assigned, for the amorphous polymers, to neutral triplet states generated by geminate recombination of bound radical pairs and, for the more crystalline polymers, to the direct observation of the geminate recombination of these bound radical pairs to ground. These observations are discussed in terms of a two-step kinetic model for charge generation in polythiophene/PCBM blend films analogous to that reported to explain the observation of exciplex-like emission in poly(p-phenylenevinylene)-based blend films. Remarkably, we find an excellent correlation between the free energy difference for charge separation (deltaG(CS)rel) and yield of the long-lived charge generation, with efficient charge generation requiring a much larger deltaG(CS)rel than that required to achieve efficient PL quenching. We suggest that this observation is consistent with a model where the excess thermal energy of the initially formed polaron pairs is necessary to overcome their Coulombic binding energy. This observation has important implications for synthetic strategies to optimize organic solar cell performance, as it implies that, at least devices based on polythiophene/PCBM blend films, a large deltaG(CS)rel (or LUMO level offset) is required to achieve efficient charge dissociation.