Short-term test-retest reliability of the 10-metre fast walk test (10mFWT) and 6-minute walk test (6MWT) was evaluated in 31 ambulatory children with cerebral palsy (CP), with subgroup analyses in Gross Motor Function Classification System (GMFCS) Levels I (n=9), II (n=8), and III (n=14). Sixteen females and 15 males participated, mean age 9 years 5 months (SD 3y 7mo, range 4y 3mo-18y 2mo). Twenty had spastic diplegia, while the others had another form of CP. Retest interval varied from 1 to 4 weeks (mean 10.6d [SD 6.4]). Intraclass correlation coefficients (ICCs) estimated reliability. The 10mFWT ICC was 0.81 (95% confidence interval [CI] 0.65-0.90) across participants, and greater than 0.59 in GMFCS subgroups (95% CI lower bound greater than 0.01). The 6MWT ICC was 0.98, and greater than 0.90 in GMFCS subgroups (95% CI lower bound greater than 0.64). Bland-Altman plots indicated bias towards higher 6MWT retest distances in GMFCS Level I. Minimum detectable change (95% CI) was 61.9, 64.0, and 47.4m for the 6MWT within GMFCS Levels I, II, and III respectively. The conclusion is that while the 10mFWT showed inadequate test-retest reliability given its wide 95% CI, the 6MWT demonstrated good to excellent reliability. Investigation of the need for a practice walk when administering the 6MWT with children in GMFCS Level I is recommended to establish their fastest pace.