Intrathecal (IT) injection of neostigmine (a cholinesterase inhibitor) has been reported to produce a significant anti-nociceptive effect in a number of inflammatory pain models. However, a potential anti-inflammatory effect of IT neostigmine in these models has not been investigated. In the present study, we have examined the 'anti-inflammatory effect of IT injection of neostigmine' (AI-NEO) using a standard mouse air pouch model by evaluating the effect of AI-NEO on zymosan-induced leukocyte migration and myeloperoxidase (MPO) release. IT neostigmine was found to suppress both leukocyte migration and MPO degranulation in a dose dependent manner. We then established which subtypes of cholinergic receptors were involved in this AI-NEO. IT pretreatment with atropine (a muscarinic receptor antagonist) but not hexamethonium (a nicotinic receptor antagonist) completely blocked the IT neostigmine anti-inflammatory effect. Subsequent experiments showed that IT pretreatment with methoctramine (a muscarinic type 2 (M2) receptor antagonist), but not pirenzepine (M1 receptor antagonist) or 4-DAMP (M3 receptor antagonist), suppressed the AI-NEO. We then evaluated whether adrenal glandular activity was important in the AI-NEO. Adrenalectomy significantly blocked the AI-NEO, while intraperitoneal pretreatment with the beta-adrenoceptor antagonist (propranolol), but not the corticosteroid antagonist (RU486) reversed AI-NEO. In conclusion, these results indicate that IT neostigmine facilitates the activation of spinal M2 receptors and this activation ultimately leads to release of adrenal catecholamines which contribute to the anti-inflammatory effect observed at the site of tissue inflammation.