1. Previous studies have provided evidence that activation of beta-adrenoceptors on cholinergic nerve terminals can inhibit neurotransmission in the airways. However, in most cases, this conclusion has been based on indirect evidence obtained from mechanical experiments where changes in airways smooth muscle tone were measured. 2. We have assessed whether modulation of cholinergic neurotransmission by beta-adrenoceptor agonists is due to a pre- or post-junctional action by investigating the effect of isoprenaline on contractile responses evoked by exogenous acetylcholine (ACh) and electrical field stimulation (EFS; 4 Hz, 40 V, 0.5 ms pulse width every 15 s), and on EFS-induced ACh release from cholinergic nerves innervating guinea-pig and human trachea. Furthermore, the subtype of beta-adrenoceptor which modulates neurotransmission and the potential role of cyclic AMP in this response were evaluated. 3. In guinea-pig trachea, isoprenaline (1 nM-1 microM) inhibited the contractile response evoked by exogenous ACh (1 microM) to a similar extent to that evoked by EFS (EC50 = 19.9 and 23 nM, respectively). 4. In epithelium-denuded guinea-pig strips treated with indomethacin (10 microM), isoprenaline significantly enhanced EFS-induced ACh release from cholinergic nerve terminals (by 36% at 0.3 microM). This effect was blocked by propranolol and ICI 118, 551 (each 0.1 microM). In contrast, isoprenaline failed to affect EFS-induced ACh release from parasympathetic nerves innervating human trachea. 5. To evaluate the role of cyclic AMP in the beta-adrenoceptor-induced facilitation of cholinergic neurotransmission, the effects of various cyclic AMP elevating drugs on ACh release were studied. Forskolin (10 microM) significantly augmented (by 17%) EFS-induced ACh release, an effect which was not reproduced by 1,9-dideoxyforskolin (10 microM) which does not activate adenylyl cyclase. Similarly, the cyclic AMP analogue, 8-bromo-cyclic AMP (1 mM) and cholera toxin (1 microgram ml-1) facilitated ACh output by 22 and 47% respectively, whereas prostaglandin E2 (PGE2, 0.1 nM-1 microM) inhibited this response (by 67% at 1 microM). 6. Zardaverine (10 microM), a dual inhibitor of the phosphodiesterase (PDE)3 and PDE4 isoenzyme families, did not affect EFS-induced ACh release and failed to facilitate the actions of either isoprenaline or PGE2. Similarly, neither SK&F 94120 (10 microM) nor rolipram (10 microM), selective inhibitors of PDE3 and PDE4 respectively, significantly affected the release of ACh in response to EFS. 7. The result of this study suggests that isoprenaline facilitates cholinergic neurotransmission in guinea-pig, but not human, trachea by activation of pre-junctional beta 2-adrenoceptors, an effect that may be mediated via activation of the cyclic AMP/cyclic AMP-dependent protein kinase cascade. Furthermore, the data presented herein illustrate the need to undertake direct measurements of neurotransmitter release when examining the effect of agents purported to act pre-junctionally.