Metal oxide sensors are widely used in the so-called "electronic noses", and it is proposed that these equipment can be used to objectify food aromas. Due to the fact that, up to now, systematic approaches to correlate the intensity of sensor responses with the structure of a given volatile have scarcely been performed [K. Suzuki, T. Takada, Highly sensitive odour sensors using various SnO2 thick films, Sans. Actuators, B 24-25 (1995) 773-776; B. Lalauze et ai., High sensitivity materials for gas detection, Sens. Actuators, B 8 (1992) 237-243; K. Fukui, Detection and measurement of odor by sintered tin oxide gas sensor, Sens. Actuators; B 5 (1991) 27-32; J.W. Gardner, A. Pike et al., Integrated array sensor for detecting organic solvents, Sens. Actuators, B 13-14 (1993) 355-357.], it is as yet not possible to predict the sensitivity and the specivity of metal oxide preparations vs. a given chemical structure. Using the SOMMSA approach [T. Hofmann et al., High resolution gas chromatography/selective odorant measurement by multisensor array (HRGC/SOMSA): a useful approach to standardize multisensor arrays for the use in the detection of key food odorants, Sens. Actuators, B 41 (1997) 81-87.], mixtures of alkanes, alcohols, aldehydes tall with chain length of 8-14 carbon atoms) and acids (chain length of 2-10 carbon atoms) were applied to different self-prepared metal oxide mixtures and the signal intensities were monitored. In addition, quantitative experiments were performed to determine the detection threshold of sensors. E.g. (E,E)-2,4-non-adienal, one of the most important odorants in fresh cucumbers, could be clearly detected at a level below 4.0 ng/ml (He) by a sensor containing a mixture of ZnO and SnO2. (C) 2000 Elsevier Science B.V. All rights reserved.