2. With these stimulation parameters, which yielded about the same fractional secretion of [3H]ACh, and with eserine (10 /LM) present in the medium, atropine (1 /M) enhanced the 'indirectly', electrically evoked secretion 3-65 + 034 (n = 6) fold, and that caused by 40 mm or 80 mM-potassium 1-82+0-06 (n = 6) or 1P55+0-09 (n = 10) fold, respectively. Atropine thus enhanced 'indirectly', electrically evoked secretion 4-fold more than that caused by 'direct' depolarization of varicosities with high potassium (P < 0 001).3. This difference is not likely to be caused by depression of the sensitivity of the presynaptic muscarinic receptors to ACh released by nerve stimulation, caused by the hypertonicity of the medium in the potassium stimulation experiments. The medium made hypertonic by addition of Tris-HEPES (80 mM) did lower the binding affinity of membrane preparations of (pre-and post-synaptic) muscarinic receptors, to carbamylcholine, and also the contractile responsiveness of the longitudinal muscle to this agent, in both cases to about one half. But it did not appear to alter the responsiveness of either pre-or post-synaptic muscarinic receptors to endogenous ACh, released by nerve stimulation.4. The results support a dual-mode model for the muscarinic negative feed-back control of ACh secretion from the nerve terminals of this preparation, mainly operating by restriction of the invasion of terminals, and only secondarily by depression of the efficiency of depolarization-secretion coupling in invaded varicosities.5. Since this model has earlier been proposed to apply for the control of secretion of [3H]noradrenaline from the micro-anatomically similar nerve terminals of P. ALBERTS, T. BARTFAI AND L. STJARNE noradrenergic nerves, the present findings suggest that the model may have a wider biological significance, and possibly apply to the control of the secretary activity of bouton8-en-pa88ant nerve terminals in general.