The responses to changes in stimulus azimuth of 220 high best-frequency (BF) (greater than 3 kHz) units in the central nucleus of the inferior colliculus of the anesthetized cat were studied with BF tones (220 units) and noise stimuli (84 units). By this means we hoped to gain some insights into the way the azimuthal locations of high BF stimuli were represented in the inferior colliculus. For each unit the discharge rate was determined for stimuli located along a plane tilted at 20 degrees above the horizontal. This plane was chosen to optimize pinna directionality. Locations in the frontal field were sampled in 10-20 degree steps around a 170 degree arc. These measurements were repeated at a number of different stimulus intensities until the directional properties of the unit became clear. Units for which the functions relating discharge rate to azimuth for a given stimulus showed a clear feature (peak or border), the azimuthal location of which varied little with intensities between 20 and 40 dB above threshold, were defined as being azimuth selective for that stimulus. Only 13% of units were azimuth selective for BF tones, whereas 44% were selective for noise. Many azimuth functions for selective units were of the plateau-shaped type for which relatively high discharge rates occurring at most contralateral azimuths declined steeply to near zero and remained low for most ipsilateral azimuths. These plateau-shaped functions were most common for tonal stimuli. Other functions showed a fixed azimuth of maximum firing (best azimuth); these were more common for noise than for tonal stimuli. Detailed azimuth functions for both tone and noise stimuli were measured for 63 units. Some exhibited the same kind of azimuth function to both stimuli. However, 18 units were azimuth selective to noise but not to tones. The borders of plateau-shaped functions obtained using both noise and tonal stimuli were concentrated within 20 degrees of the median plane. Very few units had borders that spanned peripheral ipsilateral or contralateral azimuths. Although the best azimuths of some noise azimuth functions were observed to lie at these peripheral azimuths, the majority occurred around 20 degrees contralateral to the median plane. The recording sites for units were related to a three-by-three matrix of rostrocaudal and mediolateral locations across the central nucleus. Units that were azimuth selective to noise were distributed fairly evenly throughout the central nucleus, whereas units azimuth selective to tones formed highest proportions rostrally.(ABSTRACT TRUNCATED AT 400 WORDS)