BACKGROUND Recent estimates suggest that there are over one billion people with disabilities in the world and 80% of them live in low- and middle-income countries. Community-based rehabilitation (CBR) is the strategy endorsed by the WHO and other international organisations (ILO, IDDC and others) to promote the inclusion of people with disabilities, particularly in low- and middle-income countries. The coverage of CBR is currently very low, and the evidence-base for its effectiveness needs to be assessed in consideration of scaling up of this intervention. OBJECTIVES To assess the effectiveness and cost-effectiveness of CBR for people with physical and mental disabilities in low- and middle-income countries, and/or their family, their carers, and their community. SEARCH METHODS The search for studies was not restricted by language or publication status. Searches were limited to studies published after 1976. We searched 23 electronic databases: AIM, CAB Abstract, CENTRAL, CINHAL Plus, Cochrane Database of Systematic Reviews, DARE (The Cochrane Library), EconLit, EMBASE, ERIC, Global Health, HTA Database, IBSS, IMEMR, IMSEAR, LILACS, MEDLINE, NHSEED, PAIS International, PsycINFO, The Campbell Collaboration Library of Systematic Reviews, Web of Science, WHOLIS, and WPRIM. We also searched relevant websites, contacted authors, screened the reference lists and tracked citations of included studies. The latest search for trials was in July 2012. SELECTION CRITERIA Controlled studies evaluating the impact of CBR offered to people with physical or mental disabilities and/or their family, their carers, and their community in low- and middle-income countries. The following study designs were eligible: randomised controlled trials, non-randomised controlled trials, controlled before-after studies, controlled interrupted time series studies, and economic studies. We excluded studies where CBR intervention took place only in health facilities or schools. DATA COLLECTION AND ANALYSIS Pairs of authors independently screened the search results by titles/abstracts and then by full-text, independently assessed the risk of bias, and independently extracted data. We presented standardised mean differences (SMDs) and 95% confidence intervals (CI) for continuous data and risk ratios and 95% CI for dichotomous data. We undertook meta-analysis only on outcomes extracted from studies for which the disabilities, research designs and outcome measures were agreed to be sufficiently consistent to allow pooling of data. Meta-analysis was not performed on other outcomes because the outcomes extracted from studies did not measured the same construct, the intervention was not directed at the same disability condition, or the research designs were not similar. This decision about pooling was made post-hoc and differs from the protocol. RESULTS We included 15 studies: 10 randomised controlled studies, two non-randomised controlled studies, two controlled before-after studies, and one interrupted time series study. The primary focus of 14 of the interventions was on the health component of the CBR matrix, one focused on the education component, and few included other components. Of the 15 studies, six focused on physical disabilities (stroke, arthritis, chronic obstructive pulmonary disease) and nine on mental disabilities (schizophrenia, dementia, intellectual impairment). Most of the interventions targeted both people with disabilities and their carers, although most of the studies evaluated the effect of the intervention on the person with disabilities only. Only one study focused on children as the beneficiaries of CBR. There were eight studies from East Asia and Pacific, two from South Asia, two from Europe and Central Asia, one from the Sub-Saharan Africa, one from Latin America & the Caribbean, and one from the Middle East and North Africa. The heterogeneity between studies in terms of disabilities, research designs and outcomes meant that the review relies on a narrative summary of the studies and meta-analysis was only conducted with the three studies on dementia, and only for a limited set of outcomes on users and carers. Among the six studies focusing on CBR for people with physical disabilities, two randomised controlled trials and one controlled before-after study showed a beneficial effect of the intervention for stroke on a range of outcomes while one non-randomised controlled trial found a less beneficial effect; one interrupted time series study found a beneficial impact of CBR for arthritis; and one non-randomised controlled trial showed a positive impact of CBR for people with chronic obstructive pulmonary disease. The nine studies assessing the impact of CBR for people with mental disabilities showed a beneficial effect, including: three randomised controlled trials, one non-randomised controlled trial, and one controlled before-after study on CBR for schizophrenia; three randomised controlled trials on CBR for dementia; one randomised controlled trial on CBR for intellectual disability. The dementia trials were under-powered to show a significant result, but when pooling data from the three studies, meta-analyses suggested the intervention improved carers’ clinical status (SMD=-0.37, 95% CI=-1.06-0.32) and carers’ physical quality of life (SMD=0.51, 95% CI=0.09-0.94) and carers’ social quality of life (SMD=0.54, 95% CI=0.12-5.97). However, they also suggested the intervention did not improve clinical status (SMD=0.09, 95% CI=-0.47-0.28) and quality of life (SMD=0.22, 95% CI=-0.33-0.77) of people with disabilities, carers’ burden (SMD=-0.85, 95% CI=-1.24-0.45), carers’ distress (SMD=-0.16, 95% CI=-0.54-0.22), carers’ psychological quality of life (SMD=0.11, 95% CI=-0.31-0.53), or carers’ environmental quality of life (SMD=0.07, 95% CI=-0.35-0.49). No economic evaluations meeting the inclusion criteria were found. Methodological concerns were raised about the quality of the studies. AUTHORS’ CONCLUSIONS The evidence on the effectiveness of CBR for people with disabilities in low- and middle-income countries suggests that CBR may be effective in improving the clinical outcomes and enhancing functioning and quality of life of the person with disabilities and his/her carer. However the heterogeneity of the interventions and scarcity of good-quality evidence means that we should interpret these findings with caution. More well-designed and reported randomised controlled trials are needed to build a stronger evidence-base. These studies need to be sufficiently powered, and focus on all different components of the CBR matrix and not only the health component. Furthermore, evidence is needed on a broader client groups including children, and economic evidence must be collected.