The purposes of this study were to compare the reliability and validity of an 88-item version of the Gross Motor Function Measure (GMFM-88) and a 66-item version of GMFM (GMFM-66) in children with cerebral palsy (CP). The GMFM was completed in 154 children with CP (age range = 6~12). The internal consistency of the GMFM was calculated by Cronbach's for judging reliability. The reliability of GMFM-88 and GMFM-66 were both above .99. The validity of measurement obtained by the GMFM was assessed by examining the unidimensionality of items and by comparing Gross Motor Function Classification System (GMFCS) levels with tests of the GMFM. Both the GMFM-88 and GMFM-66 were satisfied with unidimensionality. Discriminant validity was demonstrated on significant decreases in scores with increasing GMFCS levels in both measurements. However, GMFM-66 was a more sensitive discriminant in GMFCS level 1 and level 2 and in level 2 and level 4. This study reported a comparison of the reliability and validity of GMFM-88 and GMFM-66. The results of this study have implications for the information on the psychometric properties of two versions of GMFCS. This information will be useful for the selection of tools in clinics.