1. Nicotinic acetylcholine (ACh) receptor-rich patches prepared from rat myotubes were used as focal ACh detectors to record the release of ACh from magnocellular basal forebrain (MBF) neurones from 11- to 14-day-old postnatal rats maintained in dissociated cell culture. 2. An action potential generated by intracellularly stimulating the MBF cell soma through a patch electrode induced a brief (mean tau(decay), 6.3 ms) short latency (1.35-5.1 ms; median 3.1 ms) burst of nicotinic channel openings in the detector patch when the latter was positioned at discrete loci along the MBF neurites. Detected ACh concentrations ranged from approximately 480 nM to > 50 microM. Concentrations increased markedly during the first 14 days in vitro and were inversely related to response latency. 3. Sites of release were generally confined to the more proximal neurites within 100 microm of the cell body and were invariably associated with the presence of small (2-3 microm diameter) phase-dark puncta located at discrete intervals along the length of the neurites or at points where short collaterals branched from the main process. Release was never detected from the cell soma except under extreme non-physiological conditions but could occasionally be elicited from growth cones at the ends of the shorter thicker neurites in the absence of a target cell. 4. Evoked release was abolished by tetrodotoxin (0.5 microM) and by superfusing with low Ca(2+)-high Mg(2+)-containing solutions (0.25 mM Ca(2+), 5 mM Mg(2+)). Myotube patch responses were antagonized by d-tubocurarine (3 microM). 5. Muscarine (10 microM) inhibited release by 70 +/- 3% (n = 12 cells). This effect was antagonized by 100 nM methoctramine but not by 100 nM pirenzepine, indicating that it was mediated by M(2) muscarinic ACh receptors. 6. These results indicate that ACh release from the processes of magnocellular cholinergic basal forebrain neurones arises from highly specialized and discrete sites, and that it can be inhibited through activation of muscarinic receptors. It is suggested that the latter results from inhibition of presynaptic Ca(2+) channels and that it might be responsible for feedback autoinhibition of ACh release from cortical afferents of nucleus basalis neurones in vivo.