Besides the well-characterized inhibitory effect of adenosine in the gastrointestinal tract mediated by A1 receptors, we recently demonstrated that endogenously generated adenosine facilitates [3H]acetylcholine release from myenteric neurons through preferential activation of prejunctional A2A receptors. The co-existence of both receptor subtypes on cholinergic neurons prompted the question of how does adenosine discriminate between these receptors to regulate synaptic transmission in the longitudinal muscle-myenteric plexus (LM-MP) of the rat ileum. Electrical stimulation of the LM-MP increased the outflow of adenosine, inosine and hypoxanthine. Myenteric neurons seem to be the main source of endogenous adenosine, since blockade of action potentials with tetrodotoxin (1 microM) or omission of Ca2+ (plus EGTA, 1 mM) in the buffer essentially abolished nucleosides release, while adenosine outflow remained unchanged when smooth muscle contractions were prevented by nifedipine (1 microM). Inhibition of ecto-5'-nucleotidase by concanavalin A (0.1 mg ml-1) produced only a moderate decrease (approximately 25%) on adenosine accumulation in the LM-MP, indicating that the extracellular catabolism of released ATP might not be a major source of the nucleoside. Data using the acetylcholinesterase inhibitor, physiostigmine (10 microM), and several subtype-specific muscarinic receptor antagonists, 4-DAMP (100 nM), AF-DX 116 (10 microM) and muscarinic toxin-7 (1 nM), suggest that cholinergic motoneurons are endowed with muscarinic M3 autoreceptors facilitating the outflow of adenosine. Surprisingly, bath samples collected after stimulating the LM-MP exhibited a relatively high adenosine deaminase (ADA) activity (0.60+/-0.07 U ml-1), which increased in parallel with the accumulation of adenosine and its deamination products. Our findings are in keeping with the hypothesis that ADA secretion, along with a less-efficient dipyridamole-sensitive nucleoside transport system, may restrict endogenous adenosine actions to the synaptic region channelling to facilitatory A2A receptors activation. Such a local environment may also limit diffusion of exogenously added adenosine towards the active zones, as we showed that this constrain may be overcome by inhibiting ADA activity with erythro-9(2-hydroxy-3-nonyl) adenine (50 microM).