In this paper we describe magnitude and phase measurements obtained from primary single unit recordings in the cat auditory nerve. Levels range from threshold to 100 dB SPL, with frequencies from 0.1–30.0 kHz. The upper limit on the phase measurements was limited by the loss of neural phase locking at 4–5 kHz. For each unit, the frequency tuning curve (FTC) was measured by the method of Kiang and Moxon [M. C. Liberman, J. Acoust. Soc. Am.63, 442–445 (1978)] to establish the threshold frequency response of the unit. Data from several selected animals, organized by characteristic frequency (CF), are presented showing phase response, group delay, frequency tuning, and tuning slope for each CF range. The major emphasis in this paper is on the ‘‘linear’’ aspects of the data as characterized by the filter properties of the single unit response, however a number of nonlinear (level‐dependent) effects are described. Data are presented showing the phase response normalized by the chochlear microphonic (CM) recorded at the round window membrane. This normalization simplifies the phase data since it produces a constant phase slope with respect to frequency (constant group delay) for high CF units ( fCF>1 kHz) for frequencies more than one octave below their characteristic frequencies. A model of CM, as measured at the round window (RW), is presented and compared to experimental CM measurements. The CM model gives a reasonable fit to the experimental data above 500 Hz. Our interpretation of the CM normalization is that it removes driver and middle ear effects. In the model we assume that the CM is generated by the displacement of the basilar membrane near the round window recording site.