The physiologic response of first-order vestibular canal neurons, recorded within the internal auditory canal with glass microelectrodes, was studied in anesthetized cats. Neurons from all three canals were subjected to velocity trapezoidal rotations about the canal axis, and about different axes extending up to 90 degrees on either side of the canal axis in "roll" and 30 degrees on either side of "pitch." Each cell examined exhibited a spontaneous discharge and responded to constant angular acceleration in a fashion predictable from the direction of the in-plane acceleration vector and the known receptor hair cell polarization. Under conditions of prolonged constant acceleration, (5 degrees/s2 for 40 s) about 30% of the units which could be classified showed adaptation, 55% did not, and 14%, termed reverse adapting cells, demonstrated a fast rise followed by a slower, continual increase during stimulation. Secondary responses (undershoot or overshoot) were noted in most adapting neurons, but were absent in the reverse adapting group. Adapting neurons were distinguished from the nonadapting group by significantly lower resting rates, more irregular interspike-interval distributions, and greater sensitivity to acceleration. When compared with nonadapting neurons, reverse adapting cells had higher spontaneous rates, less irregular spike intervals, and higher sensitivities. The mean canal sensitivity to angular acceleration for all cells was 2 spikes . s-1/deg . s-2 (range 0.3-7.4 spikes . s-1/deg . s-2). Significant differences in mean sensitivity values between canal neurons were demonstrated, with those from the anterior being the most sensitive, followed by the posterior and horizontal canals, respectively. Time constants for all canals governing the transitory rise (or fall) in rate with constant acceleration averaged 3.8 s. Small differences in mean values were noted between canals but these were not significant. Incremental time constants were found to be slightly but significantly longer (mean = 3.9 s) than decremental time constants (mean = 3.6 s). Some cells showed different tine constants to many trials of one stimulus as well as to different levels of stimulus. Most canal unitary responses were approximately linearly related to stimulus magnitudes over the range of 2-18 degrees/s2. This being the case, the angle between the canal plane and plane of stimulus become the main determinant in the first-order neural response. Here, a linear cosine relationship descriged the three-dimentionsal unitary response curve: maximum canal response was elicited with rotation about the canal axis, while no response was evoked with rotation about an axis approximately 90 degrees to canal axis. Between these two extremes, the response of a cell was determined by the cosine of the angle between the canala axis and the axis of rotation.