Measurements of the responses to tones and clicks were made from single primary auditory fibres of the caiman. The distribution of the amplitude and phase of the fundamental component of the response rate modulation over the best frequencies of the fibres is comparable to that reported in the cat, despite the fact that the basilar membrane in caiman is only 4.5 mm long. However, much higher intensities are needed in the caiman (75-85 dB SPL) than reported in the cat (20 dB SPL) to obtain systematic distributions of the phase of the responses, probably due to the larger scatter of the phase responses in the caiman. The slopes of the phase distributions are very similar to those in cat. Single unit phase responses as a function of stimulus frequency at 85 dB SPL can be approximated by one, or in fibres with low best frequency, two straight lines. At lower intensities the deviation of the phase-frequency responses from a straight line increases as the group delay at the best frequency becomes larger. The shortest latencies of click responses are obtained with rarefaction clicks. Group delay estimates obtained from the responses to clicks and from the straight line approximations of the phase-frequency responses are related in a way expected for linear filter systems and accurately predict the measured distributions of the phase of the responses over the neural best frequency. The obtained group delays and click latencies in the caiman are very similar to those reported by other workers in the cat, the squirrel monkey and the treefrog, despite large morphological and probably functional differences of their inner ears. The click latencies are also very similar to those in the pigeon. The results are consistent with the existence of a mechanical travelling wave reported previously on the basilar membrane of the caiman, but at the same stimulus level the phase characteristic of the present single unit responses is steeper and the wave length estimates from the neural population phase distributions are shorter than those observed directly in the motion of the basilar membrane. Since the neural responses are an indirect estimate of the basilar membrane motion it cannot be decided whether the difference between neural and mechanical data is due to deterioration of the basilar membrane responses during the direct measurements or whether the basilar membrane response is sharpened by additional tuning mechanisms.