The effect of device architecture on the performance of fullerene-based organic photovoltaic (OPV) devices has not been investigated so far. Here, we performed experimental and simulation studies on fullerenebased OPVs with conventional and inverted architectures to understand the contributions from photogeneration of carriers, carrier mobility, and vertical phase separation. Charge transfer from the active layer to the bottom transport layer was studied with surface photovoltage spectroscopy (SPS) measurements. Our results show that SPS signals and spectral shape depend strongly on donor concentration in the conventional architecture but not in the inverted architecture. Additionally, conventional OPV devices show better device current density−voltage performance than inverted devices, specifically producing higher short-circuit current density. Carrier generation alone cannot explain these results. X-ray photoelectron spectroscopy results show poly(3-hexylthiophene) (P3HT) enrichment on the air surface, which should have enhanced inverted device performance. Using a solar cell capacitor simulator, we studied the active layer thickness and carrier mobility effects and established the imbalance in carrier mobilities in these devices. This carrier mobility imbalance explains the architecture dependence in the SPS results, OPV device performance, and optimal active layer thickness, while total photogeneration rate and vertical phase separation predict results inconsistent with experimental observation.