The disposition of intramuscular artemether (AM) was studied in 26 Kenyan children with cerebral malaria. Antimalarial activity determined by bioassay was compared with total plasma AM plus dihydroartemisinin (DHA) determined by high power liquid chromatography (HPLC). Therapeutic levels were achieved in most subjects (21/26) within 1 h of receiving intramuscular AM (3.2 mg/kg), with close correlation between bioassay and HPLC measurements (r = 0.706). However, there was marked inter-individual variation, antimalarial activity was undetectable in 5 subjects ('non-absorbers'), and plasma concentrations were lower in subject with respiratory distress. The 50% parasite clearance time was significantly longer in non-absorbers (mean = 13.1 h, SD = 10.8 vs. mean = 7.8 h, SD = 5.5; P = 0.013). We conclude that the bioavailability of intramuscular AM in children with severe malaria may be highly variable, particularly in the presence of respiratory distress, and may be associated with an inadequate therapeutic response.