The spectrum of anticholinergic delirium is a common complication following drug overdose. Patients with severe toxicity can have significant distress and behavioural problems that often require pharmacological management. Cholinesterase inhibitors, such as physostigmine, are effective but widespread use has been limited by concerns about safety, optimal dosing and variable supply. Case series support efficacy in reversal of anticholinergic delirium. However doses vary widely and higher doses commonly lead to cholinergic toxicity. Seizures are reported in up to 2.5% of patients and occasional cardiotoxic effects are also recorded. This article reviews the serendipitous path whereby physostigmine evolved into the preferred anticholinesterase antidote largely without any research to indicate the optimal dosing strategy. Adverse events observed in case series should be considered in the context of pharmacokinetic/pharmacodynamic studies of physostigmine which suggest a much longer latency before the maximal increase in brain acetylcholine than had been previously assumed. This would favour protocols that use lower doses and longer re-dosing intervals. We propose based on the evidence reviewed that the use of cholinesterase inhibitors should be considered in anticholinergic delirium that has not responded to non-pharmacological delirium management. The optimal risk/benefit would be with a titrated dose of 0.5 to 1 mg physostigmine (0.01-0.02 mg kg(-1) in children) with a minimum delay of 10-15 min before re-dosing. Slower onset and longer acting agents such as rivastigmine would also be logical but more research is needed to guide the appropriate dose in this setting.