The most recent results obtained on the development of selective gas sensing devices for nitrogen dioxide (NO2) monitoring associating phthalocyanine-based gas sensor and nanocarbonaceous chemical filter will be described. Since its electrical conductivity is very weakly modulated by reducing gases, copper phthalocyanine (CuPc) is a relevant semiconductor for the development of resistive sensors dedicated to oxidizing pollutants and confers to sensing devices a first level of selectivity. Ozone (O3) being the main interfering pollutant for NO2 monitoring in the context of air quality control, our approach consists in the implementation of a relevant chemical filter highly impervious to O3 and weakly reactive with NO2 placed upstream a CuPc-based sensing device. Because O3 is very reactive with carbon-carbon double bonds, different nanocarbons have been investigated as filter: activated carbons, single and multi-wall carbon nanotubes, nanofibres, nanodiscs. The filtering yields towards the target gases have been for the first time experimentally quantified. We have established the strong influence of the specific surface area of the materials which must be moderate to ensure a selective filtering. By means of graphitization and fluorination treatments, the dangling bonds, the carbonaceous matrix defects and the surface oxygenated groups (SOGs) have been identified as the most active sites of reaction. Amongst all the nanocarbons studied, a mixture of nanocones (30%) and nanodiscs (70%) has exhibited the highest filtering selectivity towards O3 and the best durability. Mechanisms of reaction are proposed and confirmed by complementary characterization techniques (Raman spectroscopy, SEM, EPR, NEXAFS). An original sensing device including carbon nanodiscs as filter and CuPc-based resistive sensor has been developed and exhibits high metrological performances in agreement with the required specifications of air quality control context: high sensitivity in the 0-200 ppb concentration range, high level of repeatability, threshold and resolution close to 10 ppb, total selectivity towards NO2, low sampling time. Strategies to enhance the durability of the filter as well as the sensing device will be finally discussed and the improvements assessed.