INTRODUCTION: Type 2 diabetes mellitus is among the foremost health challenges facing policy makers in Thailand as its prevalence has more than tripled over the last two decades, accounting for considerable death, disability and healthcare expenditure. Diabetes self-management education (DSME) programmes show promise in improving diabetes outcomes, but this is not routinely used in Thailand. This study aims to test a culturally tailored DSME model in Thailand, using a three-arm cluster randomised controlled trial comparing a nurse-led model, a peer-assisted model and standard care. We will test which model is effective and cost effective to improve cardiovascular risk and control of blood glucose among people with diabetes. METHODS AND ANALYSIS: 21 primary care units in northern Thailand will be randomised to one of three interventions, enrolling a total of 693 patients. The primary care units will be randomised (1:1:1) to participate in a culturally-tailored DSME intervention for 12 months. The three-arm trial design will compare effectiveness of nurse-led, peer-assisted (Thai village health volunteers) and standard care. The primary trial outcomes are changes in haemoglobin A1c and cardiovascular risk score. A process evaluation and cost effectiveness evaluation will be conducted to produce policy relevant guidance for the Thai Ministry of Public Health. The planned trial period will start in January 2020 and finish October 2021. ETHICS AND DISSEMINATION: Ethical approval has been obtained from Thailand and the UK. We will share our study data with other researchers, advertising via our publications and web presence. In particular, we are committed to sharing our findings and data with academic audiences in Thailand and other low-income and middle-income countries. TRIAL REGISTRATION NUMBER: NCT03938233.