Brainstem evoked response audiometry (BERA) is a useful diagnostic tool for audiological or neurotological studies but its results are greatly influenced by various test parameters including subject variables, stimulus characteristics and recording parameters. So authors performed the BERA in 10 normal persons (20 ears) to establish the effects of stimulus intensity, stimulus rate, stimulus polarity, noise masking level, filter settings, electrode placements and montages upon the wave latency and amplitude and statistically analyzed the results. 1) Among subject groups, the average of pure tone averages was 9.0 dB HL and the average of click threshold was 15.0 dB nHL. 2) The effects of noise masking levels and the site of recording electrode upon the BERA waves were negligible except that the amplitude of wave V was significantly decreased in the chin electrode. 3) The most important parameters are stimulus intensity and stimulus rate. As the intensity was decreased and the rate was increased, the wave latencies were lengthened and the amplitudes were decreased. 4) Stimulus polarity had an effect on the wave I latency but not on the interpeak latency or wave amplitude. 5) Filter settings of bandpass input had effects on the wave amplitude but not on the wave latency, interpeak latency or amplitude ratio(V/I). 6) In contralateral tracing, wave V latency was lengthened comparing with the ipsilateral tracing. With these results, we set up our own test parameters and establish the normal range of BERA in 30 normal hearing(60 ears) persons.