The experimental results reviewed in this paper strongly suggest that the molecular mechanism of N-CAM-mediated cell adhesion involves the direct interaction of N-CAM molecules on one cell with N-CAM molecules on a second cell. The rate of this aggregation has a high-order dependence on the local N-CAM concentration, and is inversely related to the sialic acid content of the N-CAM molecules involved. In accordance with their relative sialic acid concentrations, the relative rates of aggregation mediated by E and A forms of N-CAM are A-A greater than A-E greater than E-E. Further removal of sialic acid from N-CAM below the level found in the A form gives little further enhancement of aggregation. These results provide one basis upon which to interpret the modulation hypothesis (Edelman, 1983) for control of N-CAM function, i.e. the adhesive strength of N-CAM bonds in an in vitro system can be altered in a graded manner over a wide range by variations in the local surface density of N-CAM or by chemical modification of N-CAM (differential sialylation). It is important to stress that these results do not preclude the possibility of other forms of modulation of N-CAM function or the function of other molecules in cell-cell interactions. It will be much more difficult to assess the role of N-CAM and the modulation of its function on pattern formation in vivo. It is pertinent to mention, however, that recent experiments on transformed neural cells (Greenberg et al., 1984) show loss of N-CAM following transformation with accompanying loss of aggregation and increased motility of the transformed cells. Aside from the possible implications for metastasis (transformation has for the first time been shown to affect a defined CAM and alter cellular sociology), these findings are consonant with the notion that alteration of surface N-CAM affects expression of other cellular processes. Clearly additional experiments are required to define the mechanisms by which this occurs. In addition to mapping the prevalence and form of N-CAM during embryonic development, it will be necessary to pruturb both its functions and the modulation of its function using reagents that will discriminate among various forms of the molecule in the embryo itself. These problems will require elegant solutions but must be solved if completely satisfactory answers to questions about the role of N-CAM in vivo are to be obtained.