When cerebral palsy (CP) is diagnosed, most often in the firm 18 months of life, the parents want to know how severe it is and, specifically, whether their child will ever walk. Neither the connection between reflexes andearly motor skills at age 2 nor the achievement of motor milestones up to age 4 years has reliably predicted walking ability at a later age. The present investigators longitudinally analyzed patterns of gross motor development by their degree of severity in a cohort study of 657 children, ranging in age from 1 to 13 years at the outset of study, who presented the full range of clinical severity seen in children with CP. The subjects, seen at 19 regional children's ambulatory rehabilitation programs in Ontario, Canada, were followed for up to 4 years. Severity of CP was classified using the five-level Gross Motor Function Classification System (GMFCS), and function was formally assessed using the Gross Motor Function Measure (GMFM). The children in this study received treatment representative of what is generally provided in the Western world. Based on 2632 GMFM assessments, five distinctive curves of motor development were distinguished, representing significantly different rates and limits of gross motor development: level I, walks without limits but more advanced gross motor skills are limited; level II, walking is possible without aids but is limited outdoors and in the community; level III, can walk with assistive devices; level IV, patients are mobile but limited, and they are transported or use powered mobility outdoors and in the community; level V, self-mobility is markedly limited even with assistive devices. As expected, the estimated limit of development declined as the severity of impairment increased. Children at levels III through V progressed significantly more rapidly than level I children, but those in levels I and II did not differ in rate of progression. A tendency was apparent for children with lower potential for motor development to reach their limit more quickly than those with greater potential, even within the same GMFCS level. Children who perform at a higher level than expected from the average level III curve were likely to level off sooner than their peers. Caregivers must not assume that further treatment is unneeded or will be ineffective when the curve appears to level off. In general, GMFCS levels remain stable over time, permitting reliable prognostic assessments. The curves describe patterns for groups of children, and it is important to recall that within each level there are variations in motor development emanating from other aspects of children's' functional status.