The cat retina contains a number of different classes of ganglion cells, each of which has a unique set of receptive field properties. The mechanisms that underlie the functional differences among classes, however, are not well understood. All of the afferent input to retinal ganglion cells are from bipolar and amacrine cell terminals in the inner plexiform layer, suggesting that the physiological differences among cat retinal ganglion cells might be due to differences in the proportion of input that they receive from these cell types. In this study, we have combined in vivo intracellular recording and labeling with subsequent ultrastructural analysis to determine directly the patterns of synaptic input to physiologically identified X-cells in the cat retina. Our primary aim in these analyses was to determine whether retinal X-cells receive a characteristic pattern of bipolar and amacrine cell input, and further, whether the functional properties of this cell type can be related to identifiable patterns of synaptic input in the inner plexiform layer. We reconstructed the entire dendritic arbor of an OFF-center X-cell and greater than 75% of the dendritic tree of an ON-center X-cell and found that 1) both ON- and OFF-center X-cells are contacted with approximately the same frequency by bipolar and amacrine cell terminals, 2) each of these input types is distributed widely over their dendritic fields, and 3) there is no significant difference in the pattern of distribution of bipolar and amacrine cell synapses onto the dendrites of either cell type. Comparisons of the inputs to the ON- and the OFF-center cell, however, did reveal differences in the complexity of the synaptic arrangements found in association with the two neurons; a number of complex synaptic arrangements, including serial amacrine cell synapses, were found exclusively in association with the dendrites of the OFF-center X-cell. Most models of retinal X-cell receptive fields, because their visual responses share a number of features with those of bipolar cells, have attributed X-cell receptive field properties to their bipolar cell inputs. The data presented here, the first obtained from analyzing the inputs to the entire dendritic arbors of retinal X-cells, demonstrate that these retinal ganglion cells receive nearly one-half of their input from amacrine cells. These results clearly indicate that further data concerning the functional consequences of amacrine cell input are needed to understand more fully visual processing in the X-cell pathway.