Although the artemisinin-associated neurotoxicity identified in vitro and in animal studies has not been confirmed clinically, only one adult study has measured cerebrospinal fluid (CSF) concentrations after administration of conventional doses. Potential artemisinin neurotoxicity could be serious in children, especially those with meningitis and, consequently, a compromised blood-brain barrier. We measured CSF/plasma artemether and dihydroartemisinin (DHA) concentrations in 32 Papua New Guinean children with a mean age of 39 months with suspected or proven severe falciparum malaria who underwent a single lumbar puncture after intramuscular artemether administration. CSF artemether concentrations were 0 to 43.5 g/liter and CSF concentration/plasma concentration ratios were 0 to 38.1%. DHA was measurable in CSF in only two children. The seven children with meningeal inflammation (CSF white cell count > 20/mm 3 ) had higher CSF artemether concentration/plasma artemether concentration ratios than those without (median, 6.7% [interquartile ratio, 2.5 to 27.8%]% versus 0.0% [interquartile ratio, 0.0 to 2.5%]; P ؍ 0.002). Meningeal inflammation was associated with a 4.6-fold increase in the CSF artemether concentration/plasma artemether concentration ratio in a population pharmacokinetic model. These data suggest that pharmacovigilance should be heightened when intramuscular artemether is given to severely ill children with evidence of meningeal inflammation.Neurotoxicity associated with artemisinin and its derivatives has been observed in vitro and in animal studies (14, 46). Brain stem neurons are particularly susceptible (8), and the duration of drug exposure is an important determinant (8, 28). On the basis of these data, short-term peak concentrations of artesunate and its active metabolite, dihydroartemisinin (DHA), after intravenous artesunate administration are considered less potentially neurotoxic than the prolonged plasma concentrations of artemether and DHA observed after intramuscular artemether administration, especially with repeated dosing (20, 28).The relevance of in vitro and animal neurotoxicity data to humans is, however, uncertain. Animal studies have involved allometrically higher doses of artemisinin drugs given for periods longer than those recommended for the treatment of human malaria (8,20,46). There are reports of cases of artemisinin-associated neurotoxicity (15,17,22,32,39), subtle hearing loss in retrospective case-control studies (47), and a suggestion of a longer recovery from coma (48, 49) in artemether-treated patients. In contrast, substantial clinical trial and observational evidence (11,12,40), as well as data from otological (25, 31) and neuropathological (10, 21) studies, has not raised safety concerns. Nevertheless, although the mortality benefits of artemisinin drugs over alternative therapies are clear (11,12), recent reviews of artemisinin-associated neurotoxicity continue to recommend pharmacovigilance, especially in potentially high-risk groups, such as children and preg...