Strong evidence has been provided that ATP can act as a transmitter not only in smooth muscle but also in peripheral ganglia and in brain. The cloning and molecular identification of two putative ATP receptors supports the previously established pharmacological receptor classifications. This review places into perspective the evidence for ATP as a neural signalling substance by examining sites of storage, release and hydrolysis, as well as potential actions and targets. The action of ATP is related to that of the nucleoside adenosine, and the potential of additional nucleotides to function as neural messenger is examined briefly.