One of the most important inhibitory modulators of synaptic transmission in mammalian brain is adenosine. At some cholinergic terminals, adenosine is known to inhibit further release of acetyl-choline. It is unclear whether adenosine is released directly at the synapse or whether ATP is co-released with transmitter and hydro-lysed to adenosine in the synaptic cleft1–4. Methods used in the past for isolating nerve terminals have not yielded homogeneous preparations, making it impossible to determine whether sufficient ATP or adenosine is released at specific synapses for inhibition of transmitter release to occur. Immunoaffinity purification techniques5–8 have recently permitted the preparation of homogeneous populations of cholinergic nerve terminals9, which release ATP upon stimulation10. We now report that in immunoisolated cholinergic nerve terminals from the striatum synaptic ectophos-phohydrolases convert this ATP to adenosine, which inhibits further acetylcholine release, but this inhibitory effect is not seen in cortical cholinergic terminals lacking the complete ectophospho-hydrolase pathway. Therefore the differing adenosine-mediated modulation in different brain areas is controlled by the presence and activity of synaptic ectophosphohydrolases.