The behavioural responses of desert locusts, Schistocerca gregaria, to solutions of four behaviourally relevant chemicals (sodium chloride, sucrose, nicotine hydrogen tartrate and lysine glutamate) applied as droplets to the hind tarsus were analysed. All responses following within 1 s of chemical stimulation were local leg avoidance reflexes, and the probability of eliciting such a response increased in a dose-dependent manner with increasing concentration for all the chemicals tested. Chemical identity, however, critically determined the concentration threshold at which the different chemicals became an effective stimulus. For example, a 2.5 mmol l(−)(1) concentration of the secondary plant metabolite nicotine hydrogen tartrate (NHT), a potent feeding deterrent to locusts, was sufficient to evoke avoidance responses in 50 % of cases, whilst for the nutrients, sucrose and lysine glutamate, 250–500 mmol l(−)(1) of the chemical was needed to induce avoidance behaviour in 50 % of the locusts. NaCl was of intermediate effectiveness, with a 50 % response rate occurring at a concentration of approximately 50 mmol l(−)(1). The latency to the start of the response following stimulation was negatively correlated with the concentration of NaCl, but for the other chemicals concentration had no effect on latency. The duration of the avoidance behaviour decreased with increasing concentration for NaCl and more weakly for NHT, but not for the other chemicals. Adding a subthreshold concentration of sucrose to 50 mmol l(−)(1) NaCl decreased the incidence of response compared with 50 mmol l(−)(1) NaCl on its own. Experiments with other mixtures combining NaCl, sucrose and NHT indicate that the frequency and dynamics of the responses to chemical mixtures cannot be simply predicted from the responses to their individual constituents.