OBJECTIVES: To provide further evidence of the test-retest reliability and agreement of the Movement Assessment Battery for Children (M-ABC), a standardized motor test used for identification and evaluation of children with mild to moderate motor impairment. Both the stability of total test scores and classification according to specified cut-off points were examined. DESIGN AND SETTING: Children were tested on the M-ABC, three times with an interval of three weeks between each assessment. PARTICIPANTS: Thirty-three 4- and 5-year-old children with poor motor performance, 24 boys and 9 girls. RESULTS: A systematic practice effect between three consecutive testing sessions was shown. This was significant for the total score and two of the three subscores. Test-retest reliability of the total test score yielded an intraclass correlation coefficient (ICC) of 0.88. The total impairment score ranged between 0 and 40. The standard error of measurement (SEM) of this score was 2.4, resulting in a least detectable difference of 6.6 (with alpha set at 5%). The agreement over three testing sessions (kappa) was 0.72. CONCLUSIONS: The total impairment score of the M-ABC is a reliable measure for identification of mild to moderate motor impairment in young children. Repeated testing on the M-ABC at intervals of three weeks results in a systematic measurement error and is not recommended. The SEM is substantial. Follow-up of preschool children with the M-ABC as a single outcome measure is not appropriate.