Spectral analysis along with zero and standard-phase shift digital fittering were performed on evoked potentials recorded from 12 normal hearing subjects. The results indicated a progressive shifting of the mean spectral content of the ABR toward the low frequencies as the stimulus intensity was lowered. Despite this, the effects of zero-phase shift high-pass digital filtering at 100 Hz (36 dB/oct) did not significantly differ between waveforms elicited by a 75 dB nHL, 55 dB nHL, and 35 dB nHL stimulus. The major response frequency of the ABR is related to the distance between the peak (IV/V) and the following major trough (approximates one-half the response period). In waveforms where the major trough occurred before 10 msec, the use of 100 Hz, 36 dB/oct, zero-phase shift high-pass filters produced only a small reduction in response amplitude, even at low stimulus intensity levels. Waveforms which had a major trough (Na,) between 10 to 15 msec were reduced in amplitude by 100 Hz, 36 dB/oct, zero-phase shift high-pass filters (the longer period of the response energy in these waveforms corresponds to a lower energy frequency). However, this trough has a latency that prevents it from being recorded on a 10 msec time base or defined as an ABR. Based on these results, the use of zero-phase shift high-pass filters with a high-pass cutoff frequency that is equal to or less than the resolution of the time base (1/time base) appears to be a desirable method of reducing muscle artifact and other electrical contamination of the ABR. Used in this manner, the filter mainly served to eliminate baseline shift and to flatten a sloping baseline. Standard phase shift filters distorted the response and resulted in significant amplitude reduction at all stimulus settings.