The actual interest on polymer light emitting diodes (PLEDs) is based on the fact that they are easy to process, which reduces the cost of fabrication and thus opening a new branch in the electronic market – the low-cost electronics. However, these devices present a limited efficiency compared to their inorganic counterparts mainly due to the unbalanced charge injection, which reduces the fluorescence emission. One of the first strategies to improve PLEDs efficiency was using a bilayer structure composed by two polymers to improve charge injection and transport, and at the same time tune charge recombination zone to reduce the effect of the electrodes on exciton quenching. Although this is a very ingenious device architecture some of these bilayer devices showed a lower efficiency than it was expected. The reason for that is attributed to the dissolution of the first polymer layer by the solvent used for the deposition of the second polymer layer, which do not allow to create a define polymer/polymer interface. Although cross-linking the first polymer layer can solve this problem, there is not a clear understanding why the presence of a graded interface between both polymer layers can lead to a change on PLED efficiency. In order to clarify the effect of a graded polymer/polymer interface as compared to a sharp one on the functioning of a PLED, we performed computer experiments using a mesoscopic model of a bilayer PLED developed by us that considers the morphology of both polymers at nanoscale and their properties at molecular scale. The results present in this work show clearly a significant change on the charge recombination profile within the polymer device depending on the type of interface formed between the two polymers, which can be a plausible explanation for the loss of efficiency in the bilayer 7-CN-PPV/PPV LED.