This study was designed to explore a potential representation of sound azimuth in the primary auditory cortex (AI) of the cat by the relative latencies of a population of neurons. An analysis of interspike intervals (ISI) was done to asses azimuth information in the firings of the neurons after the first spike. Thus latencies of simultaneously recorded single-unit (SU) spikes and local field potentials (LFP) in AI of cats were evaluated for sound presented from nine speakers arranged horizontally in the frontal half field in a semicircular array with a radius of 55 cm and the cat's head in the center. SU poststimulus time histograms (PSTH) were made for each speaker location for a 100-ms window after noise-burst onset using 1-ms bins. PSTH peak response latencies for SUs and LFPs decreased monotonically with intensity, and most of the change occurred within 15 dB of the threshold at that particular azimuth. After correction for threshold differences, all latency-intensity functions had roughly the same shape, independent of sound azimuth. Differences with the minimum spike latency observed in an animal at each intensity were calculated for all azimuth-intensity combinations. This relative latency showed a weakly sigmoidal dependence on azimuth that was independent of intensity level >40 dB SPL. SU latency differences also were measured with respect to the latencies of the LFP triggers, simultaneously recorded on the same electrode. This difference was independent of stimulus intensity and showed a nearly linear dependence on sound azimuth. The mean differences across animals for both measures, however, were only significant between contralateral azimuths on one hand and frontal and ipsilateral azimuths on the other hand. Mean unit-LFP latency differences showed a monotonic dependence on azimuth with nearly constant variance and may provide the potential for an unbiased conversion of azimuth into neural firing times. The general trend for the modal ISI was the same as for relative spike latency: the shortest ISIs were found for contralateral azimuths (ISI usually 3 ms) and the longer ones for ipsilateral azimuths (the most frequent ISI was 4 ms, occasionally 5 ms was found). This trend was also independent of intensity level. This suggests that there is little extra information in the timing of extra spikes in addition to that found in the peak PSTH latency.