The aim of this study was to determine the reliability of family reports for the Gross Motor Function Classification System (GMFCS), a condition-specific discriminative measure of severity of movement disability for children with cerebral palsy (CP). We conducted a cross-sectional survey using a short questionnaire with families of children with CP for whom we already had ratings of GMFCS level made by a health professional. We assessed the potentially confounding effect of whether the family had discussed the GMFCS with a professional. Two hundred and one questionnaires were posted to families of which 97 (48%) were completed and returned. Mean age of the children (53 males, 40 females) was 9 years 5 months (SD 1 year 1 month), range 6 to 11 years. Children of the families who responded encompassed the spectrum of types and distribution of impairment and severity of movement disability. The intraclass correlation coefficient (ICC) of agreement between professionals and families who had discussed their child's GMFCS level with a health professional (n=35) was 0.97 (95% confidence interval [CI] 0.96 to 0.98); for those who had not (n=52) the ICC was 0.92 (95% CI 0.91 to 0.93); and for the whole sample (n=93) the ICC was 0.94 (95% CI 0.90 to 0.96). Stability between ratings made by health professionals for children when they were in the 4 to 6 year age band of the GMFCS and ratings made by families for the same children when they were in the 6 to 12 year age band (n=35) was ICC=0.96 (95% CI 0.95 to 0.97). The excellent agreement demonstrated in this study suggests that family reports of the GMFCS made by using our questionnaire provide a reliable method for measuring gross motor function in children between 6 and 12 years old. This might be more efficient for observational studies of large populations, experimental research, or community health administration than direct observation, particularly when professional assessment is not feasible.