Meeting the upcoming NOx emissions standards is a major challenge for the lean-burn engines, thus requiring a highly efficient exhaust gas aftertreatment. Currently, the Selective Catalytic Reduction (SCR) appears to be the most promising technology, especially when operated with two kinds of reductants: ammonia (generally derived from urea) and ethanol. In order to reach high conversion levels while avoiding the overinjection of the reductant, a very accurate model-based control assisted with at least one NOx sensor is required.This study focuses on the sensitivity of NOx sensors to the main nitrogenous species encountered: ammonia, isocyanic acid (HNCO) and hydrogen cyanide (HCN). The cross-sensitivity to ammonia is the only one to be already described in literature and already used in the urea-SCR control systems to limit the risks of ammonia-slip. However, HNCO can also be found downstream of a catalyst during urea-SCR if the urea delivery or the catalyst are deficient. With an ethanol-SCR system, the very dangerous HCN might be found if the selectivity of the catalyst deteriorates. Both species were found to interfere with the NOx sensor signal: the observation of the these cross-sensitivities was stated on an engine test-bench while additional characterizations were performed on a synthetic gas bench. Meanwhile, nitrous oxide, ethanol or aldehydes appeared not to bias the sensor signal.The obtained results suggest that the identified cross-sensitivities may be used to detect the presence of HNCO for urea-SCR or HCN for ethanol-SCR instead of or in addition to ammonia, due to different interference factors. These particularities are intended to help diagnose some failures of the SCR aftertreatment system.