Objectives: To trial high-dose midazolam sedation protocol for uncooperative patients with suspected psychostimulant-induced behavioural disorders. End-points were effectiveness and safety. Methods: A prospective pilot study was undertaken with a convenience sample of adult, uncooperative patients with suspected psychostimulant-induced severe behavioural disorders. The protocol was midazolam in 10 mg increments, i.m. or i.v., at 10 min intervals, up to four doses and titrated to an end-point of rousable drowsiness. Results: Sixty-two patients were enrolled. Two-thirds of the patients required only one dose of midazolam; 88% of the sample were sedated with two doses. Six and a half per cent of patients were not sedated after four doses. A Glasgow Coma Score of eight or less was prolonged in eight patients. Airway problems requiring an adjunct were present in four patients. Recent psychostimulant use was present in only 55% after full assessment. Conclusions: High-dose midazolam protocols cannot be supported as universally safe. High-dose protocols for severe behavioural disturbance are not more effective, with failures occurring even after repeated dosing.