In the present study, the test-retest variability of distortion product otoacoustic emissions (DPOAEs) due to various causes has been assessed quantitatively at individual frequencies using the ILO88/92 system. The short-term variance was generally lower than 3 dB2. Changing the position and the fit of the probe may affect the level of background noise in the ear canal, which interferes with the responses of the DPOAEs mainly at low frequencies and can also influence the interaction of the resonances of the outer ear and acoustic stimuli. The long-term variance was significantly greater than short-term variance, but did not differ significantly from variance of refitting the probe. The total DPOAE variance was small across all the frequencies above 1000 Hz except for a small peak in the 2500 Hz range. This may be related to the low response amplitudes in this region. The limitation of DPOAE detection at frequencies below 1000 Hz must lead to caution in the interpretation of results in this frequency range. The magnitude of the total variance was similar to that found in audiometric studies.