Although trihexyphenidyl is used clinically to treat both primary and secondary dystonia in children, limited evidence exists to support its effectiveness, particularly in dystonia secondary to disorders such as cerebral palsy. A prospective, open-label, multicenter pilot trial of high-dose trihexyphenidyl was conducted in 23 children aged 4 to 15 years with cerebral palsy judged to have secondary dystonia impairing function in the dominant upper extremity. All children were given trihexyphenidyl at increasing doses over a 9-week period up to a maximum of 0.75 mg/kg/d. Trihexyphenidyl was subsequently tapered off over the next 5 weeks. Objective motor assessments were performed at baseline, 9 weeks, and 15 weeks. The primary outcome measure was the Melbourne Assessment of Unilateral Upper Limb Function, tested in the dominant arm. Tolerability and safety were monitored closely throughout the trial. Of the 31 children who agreed to participate in the study, 5 failed to meet entry criteria and 3 withdrew due to nonserious adverse events (chorea, drug rash, and hyperactivity). Three children required a dosage reduction because of nonserious adverse events but continued to participate. The 23 children who completed the study showed a significant improvement in arm function at 15 weeks P = .045) but not at 9 weeks P = .985). Post hoc analysis showed that a subgroup (n = 10) with hyperkinetic dystonia (excess involuntary movements) worsened at 9 weeks (P = .04) but subsequently returned to baseline following taper of the medicine. The authors conclude that scientific evidence for the clinical use of trihexyphenidyl in cerebral palsy remains equivocal. Trihexyphenidyl may be a safe and effective for treatment for arm dystonia in some children with cerebral palsy if given sufficient time to respond to the medication. Post hoc analyses based on the type of movement disorder suggested that children with hyperkinetic forms of dystonia may worsen. A larger, randomized prospective trial stratified by the presence or absence of hyperkinetic movements is needed to confirm these results.