An experimental method is described for simulating an odor plume in the field so that its fine-scale characteristics may be determined. The odor arrived at a series of fixed points up to 15 m from the source in a series of discrete bursts, which were widely distributed in time but were typically 0.1 s long and 0.5 s apart. The strengths of the bursts were widely distributed and some contained considerable fluctuations. An odor plume is not continuous, but intermittent, and appears at a fixed point downwind of the odor source as a series of bursts of odor which are variable in strength and duration. The distribution of the length of bursts and the time between them does not greatly vary with distance from source. Although concentration decays with distance from the source, instantaneous measurement would not be a reliable guide to the mean concentration and hence to position relative to source. To obtain reliable positional information from the odor plume an insect has to average received stimuli over many seconds. More information is required about the dynamics of insects'' olfactory responses.