We investigated how natural and morphed cat vocalizations are represented in primary auditory cortex (AI). About 40% of the neurons showed time-locked responses to major peaks in the vocalization envelope, 60% only responded at the onset. Simultaneously recorded multi-unit (MU) activity of these peak-tracking neurons on separate electrodes was significantly more synchronous during stimulation than under silence. Thus, the representation of the vocalizations is likely synchronously distributed across the cortex. The sum of the responses to the low and high frequency part of the meow, with the boundary at 2.5 kHz, was larger than the neuronal response to the natural meow itself, suggesting that strong lateral inhibition is shaping the response to the natural meow. In this sense, the neurons are combination-sensitive. The frequency-tuning properties and the response to amplitude-modulated tones of the MU recordings can explain the responses to natural, and temporally and spectrally altered vocalizations. Analysis of the mutual information in the firing rate suggests that the activity of at least 95 recording sites in AI would be needed to reliably distinguish between the nine different vocalizations. This suggests that a distributed representation based on temporal stimulus aspects may be more efficient than one based on firing rate.