Background: Two-thirds of people with dementia live at home supported mainly by family carers. These carers frequently develop clinical depression or anxiety, which predicts care breakdown. We aimed to assess the clinical effectiveness (long-term reduction of depression and anxiety symptoms in family carers) and cost-effectiveness of a psychological intervention called START (STrAtegies for RelaTives). Methods: We did a randomised, parallel-group trial with masked outcome assessments in three UK mental-health services and one neurological-outpatient dementia service. We included self-identified family carers of people with dementia who had been referred in the previous year and gave support at least once per week to the person with dementia. We randomly assigned these carers, via an online computer-generated randomisation system from an independent clinical trials unit, to either START, an 8-session, manual-based coping intervention delivered by supervised psychology graduates, or treatment as usual (TAU). The primary long-term outcomes were affective symptoms (Hospital Anxiety and Depression Scale total score [HADS-T]) 2 years after randomisation and cost-effectiveness (health and social care perspectives) over 24 months. Analysis was by intention to treat, excluding carers with data missing at both 12 and 24 months. This trial is registered ISCTRN70017938. Findings: From November 4, 2009, to June 8, 2011, we recruited 260 carers. 173 carers were randomly assigned to START and 87 to TAU. Of these 260 participants, 209 (80%) were included in the clinical efficacy analysis (140 START, 69 TAU). At 24 months, compared with TAU the START group was significantly better for HADS-T (mean difference -2·58 points, 95% CI -4·26 to -0·90; p=0·003). The intervention is cost effective for both carers and patients (67% probability of cost-effectiveness at the £20 000 per QALY willingness-to-pay threshold, and 70% at the £30 000 threshold). Interpretation: START is clinically effective, improving carer mood and anxiety levels for 2 years. Carers in the control TAU group were seven times more likely to have clinically significant depression than those receiving START. START is cost effective with respect to carer and patient outcomes, and National Institute for Health and Care Excellence (NICE) thresholds. The number of people with dementia is rapidly growing, and policy frameworks assume that their families will remain the frontline providers of (unpaid) support. This cost-neutral intervention, which substantially improves family-carers' mental health and quality of life, should therefore be widely available. Funding: National Institute for Health Research Health Technology Assessment programme 08/14/06.