Background: Children of parents with mental illness have an increased risk of developing mental illness themselves throughout their lifespan. This is due to genetic factors but also environmental disadvantages during childhood associated with parental mental illness. Selective primary preventive interventions for the children are recommended to mitigate risk factors and strengthen protective factors, but large-scale, longitudinal studies are needed. This study aims to investigate the effect of the Family Talk Preventive Intervention in a cohort of children and their parents with mental illness. Methods: The study is a randomized controlled trial with 286 planned families with at least one parent with any mental illness and at least one child aged 7 to 17 years. It will be carried out in the mental healthcare system in the Capital Region of Denmark. Families will be referred from hospitals and municipalities. The children and parents will be assessed at baseline and then randomized and allocated to either the Family Talk Preventive Intervention or service as usual. The intervention group will be assigned to Family Talk Preventive Intervention, a manualized programme consisting of ~ seven sessions for the family, including psychoeducation about parental mental illness and resilience in children, stimulating dialogue between family members and creating a common family narrative. The study period for both groups will be 12 months. Follow-up assessments will be conducted after 4 months and 12 months. The primary outcomes are the children’s level of functioning, parental sense of competence and family functioning. Discussion: Given the prevalence of transgenerational transmission of mental illness, a systematic approach to prevention is needed in the mental healthcare setting. This study provides valuable knowledge on the Family Talk Preventive Intervention with a large sample size, inclusion of any parental mental illness and examination of the primary outcomes. Trial registration: ClinicalTrials.gov NCT05615324. Registered on 26 October 2022. Retrospectively registered.