Plastic photovoltaic devices offer a real potential for making solar energy economically viable. Unfortunately, bulk heterojunction (BHJ) solar cells fabricated from blends of the commonly used materials poly(3-hexylthiophene), P3HT, and phenyl-C 61 -butyric acid methyl ester, PCBM, sometimes exhibit low efficiencies even when the procedures followed often produce solar cells with efficiencies exceeding 5%. In this Letter, we show that this irreproducibility is caused by subtleties in the film processing conditions that ultimately lead to poor electron extraction from the devices. For low-performing devices, photogeneration and charge extraction with a linearly increasing voltage ramp (photo-CELIV) measurements show an order-of-magnitude difference in the effective mobilities of the electrons and holes. Atomic force microscopy (AFM) experiments reveal that the top surface of these low-performing devices is nearly pure P3HT. We argue that small variations in the solvent evaporation kinetics during spin-coating of the BHJ active layer, which are difficult to control, cause PCBM to segregate toward the bottom of the P3HT film to different extents, explaining why electron extraction from the PCBM component of the BHJ is so difficult in poorly performing devices. Finally, we show that electron extraction can be greatly improved by spin-coating a thin PCBM layer on top of the BHJ before deposition of the cathode, allowing the reproducible fabrication of high-efficiency polymer solar cells.Organic photovoltaics based on bulk-heterojunction (BHJ) composites of conjugated polymers and fullerenes have shown rapid improvement in the past few years, 1,2 with power conversion efficiencies recently surpassing 6%.3 Although facile, solution-phase fabrication is one of the greatest advantages this class of solar cells has over its inorganic-based counterparts. The behavior of polymer/fullerene devices is sensitive to small variations in processing conditions; 4,5 for example, small changes in material blend ratios, 6 single-percent variations in the composition of the solvent used for spin-coating, 7,8 and changes in postfabrication treatments such as the time and/or temperature of thermal annealing 9 all can dramatically affect device performance. Perhaps even more troublesome, there is not always good reproducibility when different groups use the same processing recipe for producing polymer/fullerene thin-film photovoltaic devices, indicating that there are still processing parameters that we have not yet either correctly identified or properly learned to control in order to consistently optimize device performance.A prime example of this lack of reproducibility can be seen in BHJ devices fabricated from blends of the commonly used materials regioregular poly(3-hexylthiophene), P3HT, and phenyl-C 61 -butyric acid methyl ester, PCBM. BHJ solar cells fabricated from these materials can have power conversion efficiencies (PCEs) exceeding 5%, 9,10 but sometimes, cells fabricated with nominally identical processing conditions can ...