Spintronics has shown a remarkable and rapid development, for example from the initial discovery of giant magnetoresistance in spin valves 1 to their ubiquity in hard-disk read heads in a relatively short time. However, the ability to fully harness electron spin as another degree of freedom in semiconductor devices has been slower to take off. One future avenue that may expand the spintronic technology base is to take advantage of the flexibility intrinsic to organic semiconductors (OSCs), where it is possible to engineer and control their electronic properties and tailor them to obtain new device concepts 2 . Here we show that we can control the spin polarization of extracted charge carriers from an OSC by the inclusion of a thin interfacial layer of polar material. The electric dipole moment brought about by this layer shifts the OSC highest occupied molecular orbital with respect to the Fermi energy of the ferromagnetic contact. This approach allows us full control of the spin band appropriate for charge-carrier extraction, opening up new spintronic device concepts for future exploitation.The development and understanding of new hybrid organic/inorganic interfaces will enable considerable progress in organic spintronics for technological purposes, including processing elements, sensors, memories and conceptually different future applications. In addition to the 'standard' spintronic applications, newly developed interfaces could bring spintronic effects to the field of organic light-emitting diodes (OLEDs), as well as in the fast progressing field of organic field-effect transistors. For example, the injection of carriers with a controlled spin state could enable the amplification of either singlet or triplet exciton states 2 leading to a significant increase in the efficiency of the electroluminescence in OLEDs. Although these considerations are conceptually straightforward, no efficiency amplification has yet been reported in the literature, despite several attempts 3 . The failure of those approaches was caused by the simple reason that light emission can be detected starting from an applied voltage of a few volts, whereas state-of-the-art spin injection in organic materials persists to a maximum of around 1 V (refs 4-6). As yet, this is unexplained. Further complications arise from the fact that various reports on working devices show a wide spread of performances for apparently similar structures, highlighting the issue of reproducibility [7][8][9] . The poor reproducibility is mainly due to the unknown interplay between processing and spin transfer performance and there is little deterministic control of the interface properties. However, it has recently been demonstrated that the insertion of a barrier