Introduction Under laboratory conditions, many plants have been shown to respond to herbivore damage by producing volatiles that attract the natural enemies of the herbivores responsible for the damage (Vinson et al., 1987; Dicke et al., 1990; Stowe et al., 1995; Turlings et al., 1995; Sabelis et al., 2001). The list of crop plants that appear to use such volatile signals includes Lima bean (Dicke et al.. In addition, some species of trees are thought to produce volatile signals that influence the searching behaviour of natural enemies (Drukker et al., 1995; Scutareanu et al., 1997) and the whole field of volatile signalling has become one of the fastest growing areas of entomological research. Much of this research is compelling. For some systems, we are beginning to understand the biochemical pathways and molecular mechanisms by which signals pass from herbivore to plant, from the site of attack to distant plant tissues, and eventually result in volatile release (Alborn et al., 1997, 2000; Frey et al., 2000; Shen et al., 2000; Turlings et al., 2000). Even for die-hard sceptics like myself, the accumulation of evidence to support the idea that plants have the machinery to manipulate the foraging of pre-dators and parasitoids by volatile emission is becoming overwhelming (Sabelis et al., 2001). If you're expecting a caveat, here it comes; the vast majority of research on herbivore-induced volatile emission has taken place under laboratory conditions. If this area of research is to realize its potential in applied entomology, we need to expand the number of studies that explore volatile release and subsequent responses under field conditions. Interactions that can be observed using Y-tubes in climate-controlled laboratories may have little relevance when exposed to the variation in temperature, precipitation, humidity, wind speed and wind direction that typify agricultural environments (Takabayashi et al., 1994). More-over, agricultural fields are moderately complex ecosystems containing a suite of biotic and abiotic factors that vary in strength in both space and time (Hunter, 2002). When challenged with root-feeding nematodes, foliar rusts, and a summer drought, does the machinery that produces volatile signals still operate? If so, does it matter a hill of (Lima) beans to levels of defoliation or (ultimately what counts) to crop yield? The good news is that an increasing number of research-ers are taking their putative signalling systems out of doors (Khan et al., 1997; De Moraes et al., 1998; Thaler, 1999; Birkett et al., 2000; Kessler & Baldwin, 2001; Ockroy et al., 2001) and these studies are a breath of fresh air in the volatile-enemy literature. Their results, of course, generate more questions than they answer, but they have taken an important first step in adding some ecology into an area of research whose application will ultimately fail or succeed as a result of ecological interactions. Here, I provide some background on studies of herbivore-induced volatile emissions, what we have learned to date from the limited number of field trials that have taken place, and consider where future research efforts might be concentrated.