The bithermal caloric test has traditionally been carried out using water but air may also be used. One of the reported disadvantages of air is the high test-retest variability (Coats et al., 1976). A recent study by Moon and Munro (1996) demonstrated that the variability could be reduced by modifying the air irrigator probe to allow greater control of probe placement and direction of air flow within the ear canal. The aim of the present study was to compare the test-retest variability of the modified air technique and the conventional water technique. Twelve normal subjects underwent four full caloric tests: twice with water and twice with air. The results show that it is possible to obtain a similar test-retest variability with the two techniques. The range for canal paresis and directional preponderance are also similar. While further research is required, the modified air technique appears to provide a reliable alternative to the traditional water technique.