The present study is focused on the implementation of a novel, low cost, urban grid of nanostructured chemresistor gas sensors for ammonia concentration ([NH(3)]) monitoring, with NH(3) being one of the main precursors of secondary fine particulate. Low-cost chemresistor gas sensors based on carbon nanotubes have been developed, their response to [NH(3)] in the 0.17-5.0 ppm range has been tested, and the devices have been properly calibrated under different relative humidity conditions in the 33-63% range. In order to improve the chemresistor selectivity towards [NH(3)], an Expert System, based on fuzzy logic and genetic algorithms, has been developed to extract the atmospheric [NH(3)] (with a sensitivity of a few ppb) from the output signal of a model chemresistor gas sensor exposed to an NO(2), NO(X) and O(3) gas mixture. The concentration of these pollutants that are known to be the most significant interfering compounds during ammonia detection with carbon nanotube gas sensors has been tracked by the ARPA monitoring network in the city of Milan and the historical dataset collected over one year has been used to train the Expert System.