It has been suggested that nasal administration of ketamine may be used to induce anaesthesia in paediatric patients. We have examined the pharmacokinetics of ketamine and norketamine after nasal administration compared with rectal and i.v. administration in young children. During halothane anaesthesia, 32 children, aged 2-9 yr, weight 10-30 kg, were allocated randomly to receive ketamine 3 mg kg-1 nasally (group IN3) or ketamine 9 mg kg-1 nasally (group IN9); ketamine 9 mg kg-1 rectally (group IR9); or ketamine 3 mg kg-1 i.v. (group IV3). Venous blood samples were obtained before and up to 360 min after administration of ketamine. Plasma concentrations of ketamine and norketamine were measured by gas liquid chromatography. Statistical comparisons were performed using ANOVA and the Kruskall-Wallis test, with P < 0.05 as significant. Mean plasma concentrations of ketamine peaked at 496 ng ml-1 in group IN3 within 20 min, 2104 ng ml-1 in group IN9 within 21 min, and 632 ng ml-1 in group IR9 within 42 min. Plasma concentrations of norketamine peaked at approximately 120 min after nasal ketamine, but appeared more rapidly after rectal administration of ketamine and were always higher than ketamine concentrations in the same situation. Calculated bioavailability was 0.50 in groups IN3 and IN9 and 0.25 in group IR9. We conclude that nasal administration of low doses of ketamine produced plasma concentrations associated with analgesia, but using high doses via the nasal route produced high plasma concentrations of ketamine similar to those that induce anaesthesia. However, the large volume of ketamine required was partly swallowed and led to an unacceptable variability of effect that precludes this route for induction of anaesthesia.