The organization of local projections within the rat primary visual cortex (area 17) was investigated by tracing fibers with HRP in in vitro brain slices. The projections from different layers showed distinct laminar patterns. Layer 4 made a strong, topographically precise, projection to lower layer 2/3; weaker projections extended laterally and terminated diffusely in layer 2/3 but also ran vertically to layers 5 and 6. The connections of lower and upper layer 2/3 were reciprocal and point-to-point. Within layer 2/3, a large number of fibers ran horizontally and terminated at variable distances from the injection site without making terminal clusters. The main output from layer 2/3 was to layer 5. The most prominent projections from the upper half of layer 5 were to layers 2/3 and 6; lower layer 5, in contrast, made wide-ranging, clustered projections to layer 1, the bottom of layer 2/3, and the top of layers 4 and 5. The patches were 130-160 micron wide and spaced apart by 230-260 micron. The main projection that arose from the superficial layer 6 terminated in layer 4 above the injection site. In contrast, lower layer 6 made clustered projections to the layer 3/4 border, extending up to 2 mm in the coronal plane. The patches were 190-220 micron wide and spaced apart by 320-390 micron. Additional projections went to the layer 5/6 border and layers 1 and 2. These results indicate that geniculocortical input is processed through interlaminar connections that are topographically precise, widespread, or patchy. These connectivity patterns suggest a role for these connections in the transformation of functional maps between layers; focused projections preserve the architecture of the layers of origin, and diverging or patchy projections rearrange this organization and form new maps in the target layers (Lund: Annu. Rev. Neurosci. 11:253-288, '88). However, only a few interlaminar connections show one of these patterns in isolation, making it difficult to assign a single function to a particular connection. We, therefore, tentatively conclude that projections terminating in layers 1-4, with the possible exception of the connection between upper layer 6 and layer 4, transform functional maps. In contrast, the topographically precise projections from upper to lower layers preserve functional maps. The specific role of these connections in the construction of receptive field properties, however, is not known.(ABSTRACT TRUNCATED AT 400 WORDS)