Semiconductor gas sensors are essentially not selective to detect a single chemical species in a gaseous mixture and also the response of the sensor in most cases is influenced by the variations of ambient humidity and temperature. One of the solutions is the analysis of the dynamic response of a single sensor with modulated temperature. For the non-linear output signal the fast Fourier transform transform was calculated. The zero-order amplitude and the phases of higher harmonics were selected. These quantities served as input data for the fuzzy model of the sensor. The hybrid fuzzy sensor model, based essentially on Takagi-Sugeno-Kang (TSK) theory, comprised a two-level optimization algorithm. The authors elaborated that algorithm, utilizing conjugate gradient and genetic algorithm methods. At the output of the fuzzy model the concentrations of ethanol with minimized influence of humidity variations were obtained. (C) 2003 Elsevier Science B.V. All rights reserved.