Cortical columns associated with barrels in layer IV of the somatosensory cortex were characterized by high-resolution 2-deoxy-D-glucose (2DG) autoradiography in freely behaving mice. The method demonstrates a more exact match between columnar labeling and cytoarchitectonic barrel boundaries than previously reported. The pattern of cortical activation seen with stimulation of a single whisker (third whisker in the middle row of large hairs--C3) was compared with the patterns from two control conditions--normal animals with all whiskers present ("positive control")--and with all large whiskers clipped ("negative control"). Two types of measurements were made from 2DG autoradiograms of tangential cortical sections: 1) labeled cells were identified by eye and tabulated with a computer, and 2) grain densities were obtained automatically with a computer-controlled microscope and image processor. We studied the fine-grained patterns of 2DG labeling in a nine-barrel grid with the C3 barrel in the center. From the analysis we draw five major conclusions. 1. Approximately 30-40% of the total number of neurons in the C3 barrel column are activated when only the C3 whisker is stimulated. This is about twice the number of neurons labeled in the C3 column when all whiskers are stimulated and about ten times the number of neurons labeled when all large whiskers are clipped. 2. There is evidence for a vertical functional organization within a barrel-related whisker column which has smaller dimensions in the tangential direction than a barrel. There are densely labeled patches within a barrel which are unique to an individual cortex. The same patchy pattern is found in the appropriate regions of sections above and below the barrels through the full thickness of the cortex. This functional arrangement could be considered to be a "minicolumn" or more likely a group of "minicolumns" (Mountcastle: In G.M. Edelman and U.B. Mountcastle (eds): The Material Brain: Cortical Organization and the Group-Selective Theory of Higher Brain Function. Cambridge: MIT Press, '78). 3. Within the stereotyped geometry of the barrel field, there is considerable individual variation in the radial labeling distribution in corresponding (homotypical) columns of different cerebral hemispheres. This result is consistent with the hypothesis that dynamic processes operate to determine the connection strengths between neural elements in somatosensory cortex. It provides a basis for testing various "connectionist" and "group selection" theories of neural organization and development.(ABSTRACT TRUNCATED AT 400 WORDS)