Background Health promotion for people with mild frailty has the potential to improve health outcomes, but such services are scarce in practice. We developed a personalised, home-based, behaviour change, health promotion intervention (HomeHealth) and assessed its clinical effectiveness and cost-effectiveness in maintaining independent functioning in activities of daily living in older adults with mild frailty. Methods This trial was an individual, multicentre, parallel-group, randomised controlled trial done in England. Participants were mainly recruited from general practices in three different areas of England (the London north Thames region, east and north Hertfordshire, and west Yorkshire). Participants were individuals residing in the community who were registered with a general practice, 65 years and older with mild frailty (scoring 5 on the CFS), with a life expectancy of more than 6 months, and with capacity to consent to participate. We excluded adults residing in nursing or care homes, those with moderate-to-severe frailty or with no frailty, those receiving palliative care, and those already case managed (eg, receiving a similar ongoing intervention from the voluntary sector or community service). Eligible participants were randomly assigned 1:1 to either the HomeHealth intervention or to treatment as usual. HomeHealth is a multidomain health promotion intervention delivered by the voluntary sector at home in six sessions over 6 months. The primary outcome was independent functioning (assessed using the modified Barthel Index [BI]) at 12 months. Outcome assessments were masked and were analysed by intention to treat using linear mixed models. Incremental costs and quality-adjusted life-years (QALYs) were calculated using seemingly unrelated regression and bootstrapping. The trial is registered on the ISRCTN registry (ISRCTN54268283). Findings We recruited 388 participants between Jan 8, 2021 and July 2, 2022 (mean age 81 years, SD 6.5; 249 (64%) of 388 were women and 139 (36%) were men). 195 participants were randomly assigned to HomeHealth and 193 to treatment as usual. Median follow-up was 363 days (IQR 356-370) in the HomeHealth group and 362 days (IQR 355-373) in the treatment-as-usual group. HomeHealth did not improve BI scores at 12 months (mean difference 0.250, 95% CI -0.932 to 1.432). HomeHealth was superior to treatment as usual with a negative point estimate for incremental costs (-796; pound 95% CI -2016 to 424) and positive point estimate for incremental QALYs (0.009, -0.021 to 0.039). There were 55 serious adverse events in the HomeHealth group and 85 in the treatment-as-usual group; none were intervention related. Interpretation HomeHealth is a safe intervention with a high probability of cost-effectiveness, driven by a reduction in unplanned hospital admissions. HomeHealth should be considered as a health promotion intervention for older people with mild frailty.