AIM To assess treatment effects of botulinum toxin type A (BoNT-A) on walking of children with leg spasticity due to cerebral palsy (CP) compared with usual care. METHOD We systematically searched the databases CINAHL, Cochrane, PEDro, EMBASE, and PubMed from July 1993 until July 2009 and additionally screened reference lists. Randomized controlled trials assessing functional outcomes on walking of children with CP were included. The intervention had to contain BoNT-A into the lower limb and be compared with usual care. The methodological quality and clinical relevance were independently assessed by two of the authors (UCR, CHGB). If statistical pooling was not feasible, we performed a best-evidence synthesis. RESULTS Eight trials were included. Trials comparing BoNT-A plus usual care or physiotherapy versus usual care or physiotherapy alone showed moderate evidence for functional outcomes at 2 to 6, 12, and 24 weeks follow-up in favour of BoNT-A. Studies comparing BoNT-A versus casting showed strong evidence for no difference in effects between these interventions. A limitation of our review was the exclusion of studies not published in English, Dutch, or German. The heterogeneity of the included studies, especially for outcome measures and follow-up assessments, prompted us to refrain from statistical pooling, whichmight also be considered a limitation. INTERPRETATION The use of BoNT-A with usual care or physiotherapy seems to improve walking of children with CP, but results should be appraised carefully owing to the limited quality of included trials.