It is know that brainstem electric response audiometry(BERA) is the most available objective audiometry in these time, but there will be an error due to false assessment of the wave form of BERA by the readers. So the author studied the reliability of BERA as evaluating the threshold difference between conventional(PTA) and brain stem response audiometry in 111 normal hearing, 46 conductive deafness and 134 sensorineural deafness ears, and following results were obtained. : 1) The threshold differences between BERA and PTA were 13.56±7.39dB in normal ears, 11.85±8.30dB in conductive, and 4.10±11.29dB in sensori-neural deafness ears. 2) The correlation coefficient was 0.83-0.87 between BERA & PTA in normal, conductive and sensori-neural deafness ears, so the author considered that there is high correlation between them. 3) It is not reasonable to differentiate deafness using latency-intensity curve of BERA. 4) PTA can be estimated with BERA but there are many risks to estimate the form and classification of deafness, so the author concluded that BERA had to be in parallel with other auditory tests to evaluate deafness.