The development of conductive polymer nanocomposite (CPC) sensors for volatile organic compounds (VOC) detection has been carried out using a spray layer by layer (LbL) process. This technique was successfully used to hierarchically structure polycarbonate-multiwall carbon nanotubes (PC-CNT) solutions into a double percolated architecture as attested by atomic force microscopy (AFM) and optical microscopy (OM). PC-CNT vapour sensing behaviour was investigated as a function of CNT content, films thickness, vapour flow and vapours solubility parameter. The response ranking A(r)(toluene)>A(r)(methanol)>A(r)(water)of PC-CNT was found to be coherent with kappa(12) Flory-Huggins interaction parameters provided that signals are normalised by analyte molecules number. Signals shape was interpreted to the light of Langmuir-Henry-Clustering (LHC) model and found to be proportional to vapour content. (C) 2009 Elsevier B.V. All rights reserved.