In most neurons, the maturation of axonal growth cones to become stable synaptic terminals is accompanied by a dramatic decline in the abundance of a major growth cone component, GAP-43. Accumulation of GAP-43 persists, however, in a minority of mature synaptic terminals. What properties of axons and their terminals are affected by these changes in GAP-43 expression? Storm and colleagues first noted that the membrane- and calmodulin-binding properties of GAP-43 (a.k.a. P-57 or neuromodulin) could allow it to sequester a large fraction of calmodulin to the submembranous regions, and to release free calmodulin in response to protein kinase C activation. Analysis of evolutionarily conserved sequences in GAP-43 indicates that these properties are central to the biological effects of the protein. If GAP-43 is presumed to inactivate bound calmodulin, the network of GAP-43 in an axon terminal could be considered a regulatable calmodulin buffer, or 'calmodulin sponge', absorbing free calmodulin and releasing it in response to activation of protein kinase C. Such a calmodulin sponge has properties that could be useful in modulating the responses of membrane and cytoskeletal assembly events to calcium signals in growth cones, and in mediating long-term potentiation of neurotransmitter release from some pre-synaptic terminals.