This communication describes the development of a protocol for the air caloric stimulus for vestibular testing. The protocol is based on matching the peak responses of air and water caloric stimuli and minimizing subject discomfort. Air stimulus temperatures of 30 and 44 degrees C are used, and these appear to minimize subject complaints of pain with the test. Composite data are presented from 16 normal subjects who were tested with the protocol. A comparison is made with their responses to a standard water protocol. Test-retest data for both air and water protocols are also presented, and these data indicate similar day-to-day variation in the two methods. On the basis of the work presented, it is concluded that the air protocol produces responses equivalent to that of standard water calorics in a group of normal subjects. Neither test was found, in the present experiments, to be particularly sensitive.