1. The ability to modify the pain threshold by the two M1-muscarinic agonists: McN-A-343 and AF-102B and by the specific M2-agonist arecaidine was examined in mice and rats by using three different noxious stimuli: chemical (writhing test), thermic (hot-plate test) and mechanical (paw pressure test). 2. In the mouse hot-plate test McN-A-343 (20-50 micrograms per mouse i.c.v.) and AF-102B (1-10 mg kg-1 i.p.) produced significant antinociception which was prevented by atropine (1 microgram per mouse i.c.v.) and by the two selective M1 antagonists: pirenzepine (0.01 micrograms per mouse i.c.v.) and dicyclomine (0.08 micrograms per mouse i.c.v. or 10 mg kg-1 i.p.) but not by the specific M2-antagonist AFDX-116 (0.1 micrograms per mouse i.c.v.), naloxone (1 mg kg-1 i.p.) or by the acetylcholine (ACh) depletor hemicholinium-3 (HC-3) (1 micrograms per mouse i.c.v.). McN-A-343 and AF-102B were able to increase the pain threshold also in the mouse acetic acid writhing test and in rat paw pressure test. These antinociceptive effects were completely prevented by dicyclomine (0.08 micrograms per mouse i.c.v. or 10 mg kg-1 i.p.) but not by AFDX-116 (0.1 microgram per mouse or rat i.c.v.). 3. In contrast with the M1-agonists, the M2-agonist arecaidine (0.1-2 micrograms per mouse or rat i.c.v.) did not induce antinociception in all three analgesic tests. However, arecaidine, at the same i.c.v. doses, was able to reduce the pain threshold in the hot-plate and paw pressure tests.4. The site of muscarinic control of the pain threshold is localized in the CNS since drugs which do not cross the blood-brain barrier such as McN-A-343, pirenzepine and arecaidine exerted their effects only if injected i.c.v.5. On the basis of the above findings and existing literature we suggest that the postsynaptic muscarinic receptors involved in antinociception belong to the M1 subtype. Nevertheless, presynaptic autoreceptors (M2 subtype) may play a role in pain regulation since they are involved in modulation of endogenous ACh release.