Background: Task sharing mental health care through integration into primary health care (PHC) is advocated as a means of narrowing the treatment gap for mental disorders in low-income countries. However, the effectiveness, acceptability, feasibility and sustainability of this service model for people with a severe mental disorder (SMD) have not been evaluated in a low-income country. Methods/Design: A randomised, controlled, non-inferiority trial will be carried out in a predominantly rural area of Ethiopia. A sample of 324 people with SMD (diagnoses of schizophrenia, schizoaffective disorder, bipolar disorder or major depressive disorder) with an ongoing need for mental health care will be recruited from 1) participants in a population-based cohort study and 2) people attending a psychiatric nurse-led out-patient clinic. The intervention is a task-sharing model of locally delivered mental health care for people with SMD integrated into PHC delivered over 18 months. Participants in the active control arm will receive the established and effective model of specialist mental health care delivered by psychiatric nurses at an out-patient clinic within a centrally located general hospital. The hypothesis is that people with SMD who receive mental health care integrated into PHC will have a non-inferior clinical outcome, defined as a mean symptom score on the Brief Psychiatric Rating Scale, expanded version, of no more than six points higher, compared to participants who receive the psychiatric nurse-led service, after 12 months. The primary outcome is change in symptom severity. Secondary outcomes are functional status, relapse, service use costs, service satisfaction, drop-out and medication adherence, nutritional status, physical health care, quality of care, medication side effects, stigma, adverse events and cost-effectiveness. Sustainability and costeffectiveness will be further evaluated at 18 months. Randomisation will be stratified by health centre catchment area using random permuted blocks. The outcome assessors and investigators will be masked to allocation status. Discussion: Evidence about the effectiveness of task sharing mental health care for people with SMD in a rural, low-income African country will inform the World Health Organisation’s mental health Gap Action Programme to scale-up mental health care globally.