Current trends in artificial nose research are strongly motivated by knowledge of biological olfactory systems, but are primarily confined to improving pattern recognition strategies for data derived from a relatively simple sensor array. Biological olfactory systems are able to discriminate weak, transient, broad-band signals ranging over a poorly-defined parameter space, and therefore outperform current artificial nose systems in several respects. A biological olfactory sense organ, the insect antenna, has been exploited in a hybrid-device biosensor. An algorithm was developed to analyze the electrophysiological responses recorded from a sensor array comprised of antennae from different species of insects. A training period during which the array was exposed to known target odors established response signatures for those odors. Subsequent odor stimuli were then classified using a forced-choice nearest neighbor technique. As odorants arrived in discrete packets in the turbulent air stream, individual sensor response events lasted less than one second, and could be classified with accuracy dependant on the differential tuning of the sensor array to the compounds being classified.