We synthesized a set of signals (clicks) for earphone delivery whose waveforms and amplitude spectra, measured at the eardrum, mimic those of sounds arriving from a free-field source. The complete stimulus set represents 1816 sound-source directions, which together surround the head to form a 'virtual acoustic space' for the cat. Virtual-space stimuli were delivered via calibrated earphones sealed into the external meatus in cats under barbiturate anesthesia. Neurons recorded in AI cortex exhibited sensitivity to the direction of sound in virtual acoustic space. The aggregation of effective sound directions formed a virtual space receptive field (VSRF). At 20 dB above minimal threshold, VSRFs fell into one of several categories based on spatial dimension and location. Most VSRFs were confined to either the contralateral (59%) or ipsilateral (10%) sound hemifield. Seven percent spanned the frontal quadrants and 16% were omnidirectional. Eight percent fit into no clear category and were termed 'complex'. The size, shape, and location of VSRFs remained stable over many hours of recording. The results are in essential agreement with free-field studies. VSRFs were found to be shaped by excitatory and inhibitory interactions of activity arriving from the two ears. Some cortical neurons were found to preserve the spectral information in the free-field sound which was generated by the acoustical properties of the head and pinna, filtered by the cochlea and transmitted by auditory nerve fibers.