A randomized controlled trial comparing: a) a combination of oral chloral hydrate and promethazine to b) a continuous intravenous midazolam infusion, for maintenance sedation in critically ill children, was carried out. The level of sedation was assessed four hourly using a specifically devized sedation scale. Forty-four children entered the study of whom two were subsequently excluded. The number of satisfactory assessments (desired and actual levels of sedation equal) was significantly greater in the chloral hydrate and promethazine group (Chi-squared P < 0.01; confidence intervals of the difference 0.06 to 0.20). The number of assessments at level 5 on the sedation scale (patient restless/ distressed) was significantly greater in the midazolam group (Chi-squared P < 0.05). The total number of satisfactory assessments in the two groups were only 61 and 48% respectively, suggesting that sedation can be considerably improved. Chloral hydrate and promethazine are more effective than midazolam as maintenance sedation in critically ill children. It is possible to prospectively study the efficacy of sedative drugs in critically ill children.