Background. Hypertension is a leading cause of cardiovascular morbidity and premature death in low- and middle-income countries (LMICs), where blood-pressure control remains poor because of suboptimal medication adherence, unhealthy lifestyle behaviours, and structural barriers to care. Theory-driven Health Belief Model (HBM) interventions have shown promise, but most treat the six HBM constructs as interchangeable predictors. Formative mixed-methods work by our group in the same population indicates that the constructs do not operate uniformly: perceived barriers and cues to action are associated with systolic blood pressure (SBP) through medication adherence, self-efficacy through pathways other than adherence, and perceived severity--though near-universally endorsed--is motivationally inert. This trial will evaluate an HBM-based, mediation-informed, PRECEDE-PROCEED-guided intervention prioritised by that evidence. Methods. This parallel-group, 1:1 cluster randomised controlled trial will assign twelve clusters in Kamalganj sub-district, north-east Bangladesh, to an HBM-based intervention or usual care (480 participants, 40 per cluster). The intervention, delivered over 12 months by trained community health workers, comprises five components: group education weighted toward barriers and self-efficacy; individual barrier-mapping and motivational counselling; enabling strategies including community blood-pressure monitoring corners, drug-supply advocacy, and cultural-practice substitution; reinforcing strategies through family engagement; and provider training on gender-perception bias. Primary outcomes are change in mean SBP and the proportion achieving controlled blood pressure at 12 months. Secondary outcomes include HBM construct scores, medication adherence (Bangladesh Medication Adherence Scale), lifestyle behaviours, and folk-remedy use, assessed at baseline, 6 and 12 months. Analyses follow intention-to-treat using mixed-effects models accounting for clustering; an exploratory longitudinal mediation analysis will test whether trial-induced change in adherence mediates trial-induced change in SBP. Discussion. This trial will provide a prospective test of a mediation-prioritised HBM intervention and, if effective, a culturally tailored, scalable model for hypertension control in comparable low-resource settings. Trial registration. This protocol is registered in ClinicalTrials.gov. Identifier: NCT07426978 (Date: 16/02/2026), Unique Protocol ID- PR-1440 (https://register.clinicaltrials.gov/prs/beta/records)