Previous studies have indicated that between 60 and 80% of a population of entomopathogenic nematodes do not infect their insect hosts at any one period in time. Two hypotheses explain this behaviour: the first that there is a subpopulation of non-infectious nematodes and the second that the non-infectious group is created by inhibitory cues derived from infected insects. Through an experimental approach with the Galleria mellonella–Steinernema feltiae system we show that both mechanisms operate together. When conditions for infection were optimized, the sum of individual infection behaviours was similar to the number infecting as a population, implying observed infection rates are driven by intrinsic mechanisms. In addition, there was evidence that an infected host released a chemical cue into the environment which inhibited subsequent levels of infection. This degree of inhibition was independent of the number of infecting nematodes. Both these mechanisms are dynamic, so the observed proportion of infectious nematodes depended heavily on the time of exposure. The implications of these findings for both the design of laboratory trials and the use of entomopathogenic nematodes in biological control are discussed.