Classification of the quality and topographical pattern of motor impairment is used to describe cerebral palsy (CP). As an adjunct to a study characterizing the quality of life and participation of school-age children with CP, initial and follow-up classification of CP were compared. A cohort of 93 children (58 males, 35 females) were initially assessed at a mean age of 2 years 6 months (SD 2y 2mo) and re-assessed at 9 years 4 months (SD 2y 2mo) with a mean interval of 6 years 6 months (SD 2y 4mo) between assessments. Sixty children had Gross Motor Function Classification System levels I-III. All but one of the children were still classified has having CP at follow-up. Type of CP documented remained constant in 67 children (72%; 95% confidence interval 62-80). Clinical factors that were statistically significant (p less than 0.05) as possible predictors of CP subtype change were original classification of a non-spastic subtype or a non-spastic quadriparetic subtype. Change in CP subtype occurs in an appreciable minority of children with CP, which is likely to reflect a combination of intrinsic and extrinsic influences. Such change may challenge efforts to monitor the effects of interventions in this population.