1. To investigate the functional significance of temporally correlated discharge between nearby cells in the visual cortex, we obtained receptive- field maps of correlated discharge for 68 cell pairs in kittens and cats. Discharge from cell pairs was measured by a single extracellular electrode. A reverse correlation procedure was used to relate neural discharge to particular stimuli within a random sequence of briefly flashed bright and dark bars. Bicellular receptive fields (BRFs) were mapped by applying reverse correlation to approximately synchronous discharge from two cells. Unicellular receptive fields (URFs) were simultaneously mapped by separately applying reverse correlation to the discharge of each cell. 2. The receptive fields of the two neurons within each pair were initially studied by varying the orientation and spatial frequency of drifting sinusoidal gratings. After these tests a random sequence of appropriately oriented bars was used to evoke discharge suitable for reverse correlation analysis. For most cell pairs, the temporal pattern or strength of correlated discharge produced by such stimulation is different from that observed with stimulation by sinusoidal gratings. This indicates that visually evoked correlated discharge between nearby cells is stimulus dependent. 3. BRFs were classified according to their pattern of spatial sensitivity into three groups that roughly correspond to the single-cell receptive-field types of the lateral geniculate nucleus (LGN; center-surround) and visual cortex (simple and complex). These classifications were compared with the receptive-field types of the single cells within each pair. LGN-type and simple-type BRFs were only seen for pairs in which at least one of the cells was simple. Conversely, complex- type BRFs were only seen for pairs in which at least one of the cells was complex. 4. Because the reverse correlation procedure can be used to characterize the spatiotemporal receptive-field structure of simple cells, we were able to compare both the spatial and temporal properties associated with the URFs and BRFs of simple cell pairs. The spatiotemporal structure of the BRF of a simple-cell pair can largely be predicted on the basis of the two URFs. Although this prediction suggests the possibility that BRFs are stimulus artifacts, a shuffle procedure, in which multiple repetitions of random sequences were presented, verifies the neural origin of BRFs. BRFs emerge from specific neural pathways and are not simply a consequence of unicellular response preferences. 5. Five measures were derived from the reverse correlation analysis of simple-cell receptive fields: width, duration, optimal spatial and temporal frequency, and optimal velocity. For each simple-cell pair, BRFs were compared with the URFs of each cell. BRFs are consistently narrower in width and shorter in duration than URFs. Despite these differences in spatial and temporal extent, BRFs do not consistently differ from URFs with respect to optimal spatial and temporal frequencies, or optimal velocities. Both the mean receptive-field width and the mean duration are significantly smaller for BRFs when compared with the means associated with URFs. 6. The BRF types seen among cell pairs of 4- and 8-wk kittens are the same as those seen in the adult. Therefore the neural circuitry necessary for the establishment of BRFs does not depend on postnatal maturation past the age of 4 wk. Additionally, the median BRF-URF width and duration differences among simple-cell pairs are similar at all age groups studied. Finally, the mean BRF width and duration are significantly smaller than the mean URI width and duration for all age groups. Thus the relationship between BRF and URF widths and durations remains essentially unchanged during postnatal development. 7. These findings suggest that correlated discharge between nearby cells could represent visual information with higher spatial and temporal resolution than is possible from the discharge of single cells. The findings provide insight into the manner by which visual information is encoded via patterns of activity distributed among multiple neurons.