Antifeeding agents are mainly organic compounds issued from the metabolic pathways of plants that interrupt food intake by phytophagous insects. These substances can be toxic but generally do not directly kill the insect; it will die by inanition. Antifeeding agents are recognized by taste receptors located on the maxilla and epipharynx. In caterpillars, the two medial and lateral sensilla, the main organs involved in chemoreception, are located on the galea. They are uniporous at their apical extremity and contain four bipolar gustatory neurones, and another one that is a mecanoreceptor. As for olfactory organs, the dendritic parts of these neurones are soaked in a sensory liquor that transports chemical stimuli from outside to the membranous receptors. Antifeeding agents belong to various classes of chemicals and it is not possible to correlate their biological activity with common structural features. Main representatives are terpenoids (drimanes, clerodanes, limonoids, agarofurans) and steroids (phytoecdysteroids). Antifeeding activities can be evaluated by the amount of treated food eaten by a caterpillar in comparison with the reference. Antifeeding agents can be classified by their feeding ratio 50: FR 50 or by their antifeedant index 50: AFI 50. These values must be calculated for each studied species, each one showing different sensitivity for the same compound. Responses of gustatory neurones to a stimulation with an antifeeding agent can be recorded at the extracellular level by electrophysiological techniques, derived from the pioneer work of Hodgson. When stimulated, these neurones respond by characteristic spikes that depend of the recognition capacity of the neurone, and many hypotheses have been formulated to explain the modalities of this recognition. The most commonly accepted is that antifeeding agents selectively stimulate a special gustatory cell: the deterrent cell. Activation of this cell is often associated with simultaneous inhibition of other gustatory cells involved in perception of phagostimulants like sugars and free aminoacids. The occurrence of the deterrent cell has been demonstrated for many phytophagous caterpillars and can explain repellence for the non-host plants. Phytoecdysteroids are secondary metabolites widely spread in the plant kingdom and it has been hypothesized that they act as defensive substances against herbivorous. Two categories of caterpillars in relation to their sensitivity to the substance have been proposed: the tolerant ones that accept relatively large amounts of phytoecdysteroids in their diet, and the sensitive ones that reject diet with even low amounts. According to this classification, polyphagous species must be tolerant, and at the opposite monophagous species sensitive (diet breadth hypothesis). To verify this hypothesis, the antifeeding activity of three phytoecdysteroids (ecdysone: E, 20-hydroxyecdysone: 20E, and ponasterone: A ponA) has been evaluated on four caterpillars species, one monophagous Bombyx mori, one oligophagous Mamestra brassicae, and two polyphagous Spodoptera littoralis and Ostrinia nubilalis. Using feeding tests, it has been found that the two last instar larvae of B. mori and M. brassicae avoid diet containing 20E or ponA with a respective threshold of 200 and 500 ppm but are both insensitive to E even at higher concentrations. S. littoralis larvae (L4) are tolerant to the three phytoecdysteroids, but surprisingly O. nubilalis larvae (L5), a polyphagous species, is very sensitive, even at very low concentration (25 ppm). Electrophysiological recordings are consistent with the occurrence of the deterrent cell located in the medial sensilla of B. mori and the lateral one of M. brassicae that answers to 20E and ponA but not to E, revealing the importance of the 20-hydroxyl group on the lateral chain of the steroid nucleus. This deterrent cell is also present in both sensilla of O. nubilalis and responds to the three phytoecdysteroids at very low thresholds. Responses of S. littoralis are not different to the standard (KCl or fructose). The diet breadth hypothesis is verified for B. mori, M. brassicae and S. littoralis. It is not true for O. nubilalis, a polyphagous species, which must be tolerant to phytoecdysteroids acordingly to the hypothesis. The question is to know whether O. nubilalis is an exception, or if the diet breadth hypothesis that has been proposed after observations on a few number of species, is applicable on a large scale.