In some circumstances, a high degree of sedation that results in a child being unconscious at the time of parental separation is desirable. We set out to investigate the efficacy and safety of a rectal premedication regimen designed to produce this increased level of sedation. Sixty-seven children aged two to 24 months were randomised into two groups. Group MK received 4 mg x kg(-1) ketamine, 0.5 mg x kg(-1) midazolam and 0.02 mg x kg(-1) atropine and group MKK received 8 mg x kg(-1) ketamine, 0.5 mg x kg(-1) midazolam and 0.02 mg x kg(-1) atropine per rectum. The sedation score at the time of parental separation 30 minutes after drug administration and the response to intravenous cannulation were evaluated on a four-point scale. Respiratory rate, heart rate and arterial oxygen saturation were recorded immediately before parental separation. More patients in group MKK were asleep during separation (62 vs 35%, P < 0.05). Fewer patients in group MKK cried during intravenous cannulation (37 vs 68%, P < 0.05). Sedation scores were significantly increased at both time points. There was no difference between groups in vital signs at the time of parental separation and no adverse respiratory events occurred during the study period. In cases where a high degree of sedation following premedication in infants and toddlers is desired, the addition of 8 mg x kg(-1) ketamine to 0.5 mg x kg(-1) midazolam and 0.02 mg x kg(-1) atropine administered rectally is more efficacious than 4 mg x kg(-1) ketamine.