Objective: To summarise the effect of upper limb rehabilitation interventions on motor function in adults with traumatic brain injury. Data sources: Databases were last searched on August 2, 2008. Sources included the Cochrane Central Register of Controlled Trials (CENTRAL); Cochrane Database of Systematic Reviews, the Database of Effectiveness Reviews; MEDLINE, EMBASE, CINAHL, Physiotherapy Evidence Database (PEDro), Occupational Therapy Systematic Evaluation of Evidence database (OTseeker); Google Scholar; and reference lists of included studies. Review methods: Two reviewers determined whether retrieved abstracts met the inclusion criteria: systematic reviews and randomised controlled trials (RCTs); English language; adult participants; ≥ 50% of study participants with a brain injury; interventions designed to improve upper limb motor function. Included papers were appraised for: study design, participants, therapy approach, therapy protocol (indications, contra-indications, intensity and duration), safety and adverse events, and outcomes. The methodological quality of RCTs was rated using the PEDro scale (1–10 highest). Methodology of systematic reviews was rated using the QUOROM criteria. Results: Of the 333 references identified, six were appraised: three systematic reviews and three RCTs. Methodological quality was high for two RCTs, and moderate for one, based on the PEDro scale score. Interventions included upper limb casting, electrical stimulation, and coordination training using meal preparation tasks (making a sandwich and hot drink). In the latter trial, practice of functional kitchen tasks improved fine motor coordination speed on one of four Jebsen-Taylor hand function subtests by 9.38 seconds (95% CI, 1.1 to 17.7). Remaining trials reported non-significant effects for hand function. Small sample sizes and limited reporting of results reduce the interpretability of two RCTs. Conclusion: No conclusive evidence was found on which to base upper limb motor rehabilitation after brain injury, however, lack of evidence does not equate to evidence of no effect.