Aims: To examine test–retest reproducibility of the Assessment of Motor and Process Skills (AMPS) in children aged 8–16 years with unilateral cerebral palsy (UCP). Methods: Thirty children with mild to moderate UCP (mean age = 11y 7m, SD 2y 4m; males = 18; Manual Ability Classification System level I = 10, II = 20; Gross Motor Function Classification System level I = 9, II = 21) enrolled in a large randomized controlled trial were recruited via consecutive series sampling. Children carried out two AMPS tasks over two consecutive days according to standardized AMPS administration procedures. The standard error of measurement (SEM), smallest detectable change (SDC), 95% limits of agreement using the Bland–Altman method, and intraclass correlation coefficients (ICC; 2,1) were calculated. Results: The SDC was 0.23 logits for the AMPS motor scale and 0.30 logits for the AMPS process scale. Test–retest reliability was excellent for both the AMPS motor scale (ICC = 0.93) and the AMPS process scale (ICC = 0.86). Intra-rater reliability (n = 10) was excellent for AMPS motor scale (ICC = 0.96) and AMPS process scale (ICC = 0.98). Conclusions: The AMPS can be used by therapists with 8- to 16-year-old children with UCP as an outcome measure with changes in scores reflecting real changes in performance or capacity.