1. Micro-electrode recordings were obtained from over 100 single fibres in the cochlear nerve of the pentobarbitone or urethane anaesthetized guinea-pig. The acoustic system was calibrated at the tympanic membrane and threshold sound level measurements so corrected.2. The minimum thresholds of the fibres approached with 10-20 dB of the behavioural thresholds reported in the literature. Exceptions to this were fibres from preparations where there was evidence of malfunction of the cochlea either from abnormally low perfusion or local damage, and a few high frequency fibres. With these high threshold fibres excepted, the range of thresholds at a given frequency in any one animal was less than 20 dB.3. The slopes of the low and high frequency cut-offs of the frequency-threshold curves (;tuning curves') within 25 dB of minimum threshold, ranged from 10 to 60 and from 20 to 125 dB/octave respectively for fibres with characteristic frequencies below 2 kHz, increasing to 90-180 and 200-600 dB/octave respectively for fibres with characteristic frequencies at about 8 kHz. These slopes represent the minimum values for the high-frequency cut-offs, which increase towards 1000 dB per octave in some cases at higher levels above threshold. At 30-50 dB above threshold, the low frequency cut-offs become suddenly less steep and approximate to 5 dB per octave.4. The relative sharpness of the frequency-threshold curves, measured as the ;Q(10 dB)', i.e. the ratio of characteristic frequency to the band width at 10 dB above minimum threshold, ranged from 1 to 4 for fibres with characteristic frequencies below 2 kHz, to 3-15 for fibres with characteristic frequencies near 10 kHz.5. The slopes and ;Q(10 dB)' measures of the frequency-threshold curves of most of the abnormally high threshold fibres approximated to, or were lower than those of analogous measurements of the guinea-pig basilar membrane vibration patterns.6. Four fifths of the cochlear nerve fibres had spontaneous discharge rates greater than 1/sec. No consistent relationship was observed between the rate of this activity and response properties, with the exception that nearly half of the high threshold fibres were silent. In these and other respects the response properties to tonal and click stimuli resembled those of cochlear nerve fibres in the cat. In no case was inhibition of the spontaneous discharge by single tones observed.7. It is concluded that, contrary to earlier reports, the cochlear nerve fibres of the guinea-pig are substantially more frequency selective than the existing measurements of the guinea-pig basilar membrane displacement. In terms of band width, this discrepancy approaches a factor of ten. The finding of a considerable range of band widths within optimal preparations, and frequency-threshold curves approximating to the mechanical functions in fibres from pathological cochleas, provides circumstantial evidence for a physiologically vulnerable sharpening mechanism occurring within the cochlea subsequent to the displacement pattern of the basilar membrane.