To assess the effectiveness and cost-effectiveness of breaks in care in improving the well-being of informal carers of frail and disabled older people living in the community and to identify carer needs and barriers to uptake of respite services. Major electronic databases were searched from the earliest possible date to April 2008. Selected studies were assessed and subjected to extraction of numerical data for meta-analysis of quantitative studies and extraction of text for thematic analysis of qualitative studies. Quality of the studies was assessed using checklists specifically designed for the current review. In total, 104 papers were identified for inclusion in the quantitative synthesis, 16 of which were appropriate for meta-analysis. Carer burden was reduced at 2-6 months' follow-up in single-sample studies but not in randomised controlled trials (RCTs) and quasi-experimental studies. Depression was reduced in RCTs in the short term and for home care but not for day care. These effects, however, were not significant in random-effects models. There was a trend for longer interventions to have more positive effects than shorter interventions. There was no effect of respite on anxiety, but it had positive effects on morale and anger and hostility. Single-group studies suggested that quality of life was worse after respite use. There were increased rates of institutionalisation after respite use; however, this does not establish a causal relationship as it may be a result of respite being provided late in the caregiving career. A total of 70 papers were identified for inclusion in the qualitative synthesis. Uptake of respite care was influenced by: carer attitudes to caring and respite provision; the caregiving relationship; knowledge of, and availability of, services; the acceptability to, and impact of respite care on, care recipients; hassles resulting from the use of respite care; quality of respite care; and the appropriateness and flexibility of service provision. Carers expressed needs for active information provision about services, support offered early in the caregiving career, access to a variety of services with flexible provision, reliable transport services, continuity of care, good-quality care, appropriate environments, care that provides benefits for care recipients (socialisation and stimulation), and appropriate activities for care recipients' levels of abilities and interests. There was some evidence to support respite having a positive effect on carers but the evidence was limited and weak. It is difficult, therefore, to make recommendations as to the most appropriate form of delivery of respite, apart from the suggestion that a range of services is probably most appropriate, to provide flexibility of respite provision and responsiveness to carer and care recipient characteristics and needs and also changes in those needs over time. There is a need for further high-quality larger trials that include economic evaluations.