The development of a linear filter to optimise the signal-to-noise ratio of the auditory cortical evoked potential is described. The filter characteristics were derived from the frequency spectra of cortical potentials taken from 40 normal and 20 hearing impaired adult ears, at two test frequencies (120 tests). The performance of the filter was compared with a typical filter used in clinical practice (1.5 Hz to 15 Hz second-order Butterworth filter). Results showed that the filter produced an average increase in the signal-to-noise ratio of approximately 38%. Further comparisons were made using 14 different Butterworth filters (all second order) and the best of these, the 5 Hz to 9 Hz filter, produced a 28% improvement in the signal-to-noise ratio. The signal-to-noise ratio was calculated by comparing the absolute integral area of the average post-stimulus data to that of the pre-stimulus data. This improvement in the signal-to-noise ratio enhances signals for objective machine scoring analysis or alternatively, allows for a reduction in the number of sweeps (and hence time) required to record the evoked potential.