BACKGROUND: There is considerable literature on managing depression, burden and psychological morbidity in caregivers of people with dementia (CG). Anxiety has been a relatively neglected outcome measure but may require specific interventions. OBJECTIVE: To synthesise evidence regarding interventions that reduce anxiety in CGs. METHODS: Twenty-four studies met our inclusion criteria. We rated the methodology of studies, and awarded grades of recommendation (GR) for each type of intervention according to Centre for Evidence Based Medicine guidelines, from A (highest level of evidence) to D. RESULTS: Anxiety level was the primary outcome measure in only one study and no studies were predicated on a power calculation for anxiety level. There was little evidence of efficacy for any intervention. The only RCT to report significantly reduced anxiety involved a CBT and relaxation-based intervention specifically devised to treat anxiety, and there was preliminary evidence (no randomised controlled trials) that caregiver groups involving yoga and relaxation without CBT were effective. There was grade B evidence that behavioural management, exercise therapies and respite were ineffective. LIMITATIONS: Many interventions were heterogeneous, so there is some overlap between groups. Lack of evidence of efficacy is not evidence of lack of efficacy. CONCLUSIONS: CBT and other therapies developed primarily to target depression did not effectively treat anxiety. Good RCTs are needed to specifically target anxiety which might include relaxation techniques. Some of the interventions focussed on reducing contact with the care recipients but caregivers may want to cope with caring and preliminary evidence suggests strategies to help CGs manage caring demands may be more effective.