The generation of specific patterns of neuronal connections has usually been regarded as a central problem in neurobiology. The prevailing view for many years has been that these connections are established by complementary recognition molecules on the pre- and postsynaptic cells (the chemoaffinity theory). Experimental results obtained in the past decade, however, indicate that the view that axon guidance and synaptogenesis proceed according to restrictive chemical markers is too narrow. Although a more rigid plan may prevail in some invertebrates, the formation of specific connections in vertebrates also involves competition between axon terminals, trophic feedback between pre- and postsynaptic cells, and modification of connections by functional activity.