Grasshopper mechanosensory hair neurones respond to displacement of their associated hairs in a temperature sensitive manner: comparable increases in the number of spikes per stimulus result from increases in temperature with constant stimulus strengths and from increasing stimulus strengths at constant temperature. It is therefore not obvious that neurones in the CNS which receive inputs from mechanosensory hairs would be able to distinguish between these two parameters. The temperatures which populations of mechanosensory hairs on the thorax, head and tarsus experienced were measured in freely moving animals. Animals in thermally heterogeneous environments spent 90% of the accounted time in locations where thoracic temperatures of 32-44 degrees C were maintained (the behaviourally 'preferred' range). Head temperatures covered a wider range, and tarsal temperatures the widest. Different populations of mechanosensory hair neurones exhibited different sensitivities to temperature. Thoracic hair neurones were significantly more temperature sensitive than one of the two populations of head hairs studied, and tarsal hairs exhibited a pronounced temperature compensation in the behaviourally 'preferred' range. Wind sensitive head hairs, however, showed exceptionally high temperature sensitivities. There is a negative correlation between the temperature sensitivity of a population of mechanosensory hair neurones and the temperature variability to which those neurones are normally exposed. Implications of this correlation for the central interpretation of mechanosensory input are considered.