The aim of this study was to explore motor development in children with cerebral palsy (CP) using developmental curves for CP, subtypes, and the five severity levels of the Gross Motor Function Classification System (GMFCS). The Gross Motor Function Measure (GMFM) and the GMFCS were applied to 317 children (145 females, 172 males) with CP, aged between 1 and 15 years. The CP type distribution was spastic diplegia in 157 (49%), spastic hemiplegia in 101 (33%), spastic tetraplegia in 11 (3%), dyskinesia in 38 (12%), and ataxia in 10 (3%). Forty-five physiotherapists were trained in the GMFM and intra- and interrater reliability was tested. The GMFM was measured prospectively every 6 months up to the age of 4 years and once a year thereafter. Developmental curves were constructed for 258 children with spastic CP. About three-quarters of the children at GMFCS Level I reached 90% of the maximum GMFM score at 5 years of age. The performance peaked at 7 years of age. Children at GMFCS Level II reached 90% at a median age of 5 years, which was also the upper limit, reached by about three-quarters at 7 years of age. The majority of children at GMFCS Level III reached 80% of the GMFM by 7 years of age and most of the children at GMFCS Level IV reached 30% at 5 years and remained there. The median score for children at GMFCS Level V was 20%. The intra- and interrater reliability for the GMFM 88 among physiotherapists were Spearman's rank correlation coefficient 0.91 and 0.99 respectively. There were 931 measurements with a median of 2 (1-11) per child. The gross motor development was demonstrated for the five GMFCS levels in children with spastic CP. These kind of curves may be useful for monitoring and predicting motor development, for planning treatment, and for evaluating outcome after interventions.