The importance of Cu for CdTe solar cell absorber doping has been increasingly recognized in recent years. Currently different models are being discussed how to understand the case of Cu Cd substitutional doping in polycrystalline CdTe solar cells. In this work, an understanding is developed, which is based on a low concentration deep acceptor doped CdTe layer (N a $ 5 Â 10 14 cm À3 ,E a $ 300 meV above the valence band). Despite their non-shallow nature, Cu Cd acceptors are fully or at least heavily (>30%) ionized. The low hole concentration in CdTe ($1 Â 10 14 cm À3) originates directly from low Cu solubility in CdTe bulk material and is not caused by partial ionization or compensation as proposed by earlier models. The three to four orders of magnitude difference between bulk acceptor concentration and average Cu concentration in polycrystalline CdTe is attributed to grain boundary segregation of Cu. Our model is derived from substrate and superstrate CdTe solar cell measurements, controlled CdTe doping and quenching, Hall Effect measurements of CdTe films, numerical and analytical calculations, and a broad literature survey. Based on these results, routes to improve the conversion efficiency of CdTe solar cells are discussed. V