Abstract Background The expected increase of dementia prevalence in the coming decades will mainly be within low-and middle-income countries and those with low socioeconomic status (SES) in high income countries. Forty percent of dementia is associated with modifiable, lifestyle-related cardiovascular risk factors. Mobile health interventions with remote coaching can help reach underserved high-risk populations globally. Methods This open label, blinded endpoint, hybrid effectiveness-implementation randomised controlled trial (RCT) investigated whether a coach-supported mHealth intervention can reduce dementia risk in persons aged 55-75 years of low SES in UK or from the general population in China with at least two dementia risk factors. The primary effectiveness outcome was change in CAIDE dementia risk score (including age, education, sex, systolic blood pressure, total cholesterol, BMI, physical exercise) after 12-18 months of intervention. Main implementation outcomes were coverage, acceptability, appropriateness, adoption, feasibility, sustainability, fidelity, and costs using a mixed methods approach. Findings Between 15th January 2021 and 18th April 2023, 1,488 persons were included and randomised (734 intervention, 754 control), with 1,229 (82·6%) available for analysis of the primary effectiveness outcome. After a mean follow-up of 16 months, the mean CAIDE score improved by 0·16 points in the intervention group vs. 0·01 in the control group (Mean Difference (MD) 0·16, 95% Confidence Interval (CI) 0·03-0·29). The effect was larger in those who adhered to the intervention (MD 0·27, 95%CI 0·12-0·44) and those planning to change lifestyle (MD 0·33, 95%CI 0·10-0·57). Ten percent of those invited participated; of these, 81% adopted the intervention and 50% continued active participation throughout the study. For the individual components of the CAIDE score, there were no statistically significant differences for systolic blood pressure, total cholesterol and BMI, but in the intervention group physical inactivity was reduced more than in the control group (MD 6·7%, 95%CI 2·1 to 11·2) and more participants quit smoking (MD -3·1%, 95%CI -4·8 to -1·5). Interpretation A coach-supported mHealth intervention to improve lifestyle is modestly effective in reducing dementia risk factors in those with low SES in the UK and any SES in China. Implementation is challenging in these populations, but those reached actively participated. Whether this will result in less cognitive decline and dementia requires a larger RCT with long follow-up.