Plants that are damaged by herbivorous arthropods provide carnivorous enemies of the herbivores with important information. They emit an induced volatile blend that is highly detectable to the carnivores from a distance. Such detectable signals that indicate herbivore presence are important for the carnivores because herbivores themselves are under strong selection not to expose themselves. In addition, carnivores would benefit from a specificity of the induced plant volatiles. Whether herbivore-induced plant volatiles are reliable indicators of herbivore identity, however, has not been resolved unambiguously. Some studies support the reliability of herbivore-induced plant volatiles, while others do not. Different approaches have been used such as chemical analysis, behavioural analysis or a combination of the two. Based on the total of chemical studies one might conclude that in most cases herbivore-induced plant volatiles are not very specific for the herbivore that damages the plant. However, arthropod chemosensors are much more sensitive than the detectors of analytical instruments. Therefore, chemical analyses are not suitable to demonstrate whether or not herbivore-induced plant volatiles are reliable indicators of herbivore identity to carnivores. Behavioural studies should provide this information. In analysing carnivore behaviour it should be realised, however, that arthropod behaviour can be highly variable. Arthropod foraging decisions are affected by external and internal factors such as (a) abiotic environmental factors, (b) presence of competitors or enemies, (c) deprivation of food or oviposition sites, (d) specific deprivation of certain nutrients or (e) learning. In this paper their effect on discrimination of carnivores between volatile blends emitted by plants infested by different herbivores is reviewed. This provides testable hypotheses of why discrimination was not found in some studies. The ability of carnivores to discriminate is likely to be more common than is clear to date, which should invoke functional studies of the conditions that influence the occurrence of this discrimination.