AIM The aims of this study were to (1) determine whether an instrumented measure will reduce measurement error to less than 5? in children with cerebral palsy (CP), (2) determine agreement and reliability of this instrumented measure in both conscious and anaesthetized participants, and (3) compare the method with previously reported measures. METHOD Thirty-four ambulant children (15 males, 19 females), aged 3 to 9 years, with spastic CP were studied in a tertiary-care paediatric hospital (21 with hemiplegia, 11 with diplegia, and two with quadriplegia). The majority of children functioned at Gross Motor Function Classification System level I (n=11) or II (n=18), with five children at level III. Ankle dorsiflexion at 50 bodyweight was photographed and measured. Each child was measured when conscious and when under mask anaesthesia by two experienced assessors. RESULTS The standard error of measurement (SEM) ranged from 3.9? (anaesthetized; 95 confidence interval [CI] 3.3?4.0?) to 6.7? (conscious; 95 CI 5.3?8.0?). This compared favourably with previously reported dorsiflexion measures (SEM range 6.5?7.8?) in conscious children with CP. Intrarater reliability was good in both conditions (intraclass correlation coefficient [ICC]: range 0.95 [anaesthetized; 95 CI 0.92?0.98] to 0.86 [conscious; 95 CI 0.76?0.95]). The ICC for interrater reliability ranged from 0.87 (anaesthetized; 95 CI 0.81?0.93) to 0.65 (conscious; 95 CI 0.50?0.81). INTERPRETATION Passive ankle dorsiflexion using an instrumented measure has face validity and may assist in the improvement of reproducibility under anaesthesia for clinical research. When an individual is conscious, this technique is not better than trained assessors using conventional goniometry reported in the literature and is not recommended for routine clinical use.