A striking feature of the mature central nervous system is the precision of
the synaptic circuitry. In contemplating the mature circuitry, it is impos
sible to-imagine how more than 20 billion neurons in the human brain become
precisely connected through trillions of synapses. Remarkably, much of the
final wiring can be established in the absence of neural activity or exper
ience; so the algorithms that allow precise connectivity must be encoded la
rgely by the genetic programme. This programme, honed over nearly one billi
on years of evolution, generates networks with the flexibility to respond t
o a wide range of physiological challenges. There are several contemporary
models of how synapse specificity is achieved, many of them proposed before
the identification of guidance or recognition molecules. Here we review a
selection of models as frameworks for defining the nature and complexity of
synaptogenesis, and evaluate their validity in view of progress made in id
entifying the molecular underpinnings of axon guidance, targeting and synap
se formation.