The effects of muscarinic receptor antagonists on responses to electrical stimulation of the chorda-lingual nerve were determined in pentobarbitone-anaesthetized sheep and correlated to the morphology of tissue specimens. Stimulation at 2 Hz continuously, or in bursts of 1 s at 20 Hz every 10 s, for 10 min induced similar submandibular fluid responses (19±3 vs. 21±3 µl.min^-1 g gland^-1), while vasodilatation was greater during stimulation in bursts (-52±4 vs. -43±5 %; P<0.01). Continuous stimulation at 8 Hz induced substantially greater responses (66±9 µl.min^-1.g gland^-1 and -77±3 %). While atropine (0.5 mg.kg^-1 I.V.) abolished the secretory response at 2 and 20 (1:10 s) Hz, a small response persisted at 8 Hz (<5 %). The "M1-selective" antagonist pirenzepine (40 µg.kg^-1 I.V.) reduced the fluid response at all frequencies tested (P < 0.05-0.01), most conspicuously at 2 Hz (reduced by 69 %). Methoctramine ("M2/M4-selective"; 100 µg.kg^-1 IV; n=5) had no effect on fluid or the vascular responses but increased the protein output at 2 (+90 %, P < 0.05) and 8 Hz (+45 %, P <0.05). The immunoblotting showed distinct bands for muscarinic M1, M3, M4 and M5 receptors, and immunohistochemistry showed muscarinic M1 and M3 receptors to occur in the parenchyma. Thus, muscarinic M1 receptors contribute to the secretory response to parasympathetic stimulation but have little effect on the vasodilatation in the ovine submandibular gland. Increased transmitter release caused by blockade of neuronal inhibitory receptors of the M4 subtype would explain the increase in protein output.