IntroductionCow’s milk is a dietary staple for children in North America. Though clinical guidelines suggest children transition from whole (3.25% fat) milk to reduced (1% or 2%) fat milk at age 2 years, recent epidemiological evidence supports a link between whole milk consumption and lower adiposity in children. The purpose of this trial is to determine which milk fat recommendation minimises excess adiposity and optimises child nutrition and growth.Methods and analysisCow’s Milk Fat Obesity pRevention Trial will be a pragmatic, superiority, parallel group randomised controlled trial involving children receiving routine healthcare aged 2 to 4–5 years who are participating in the TARGet Kids! practice-based research network in Toronto, Canada. Children (n=534) will be randomised to receive one of two interventions: (1) a recommendation to consume whole milk or (2) a recommendation to consume reduced (1%) fat milk. The primary outcome is adiposity measured by body mass index z-score and waist circumference z-score; secondary outcomes will be cognitive development (using the Ages and Stages Questionnaire), vitamin D stores, cardiometabolic health (glucose, high-sensitivity C-reactive protein, non-high density lipoprotein (non-HDL), low density lipoprotein (LDL), triglyceride, HDL and total cholesterol, insulin and diastolic and systolic blood pressure), sugary beverage and total energy intake (measured by 24 hours dietary recall) and cost effectiveness. Outcomes will be measured 24 months postrandomisation and compared using analysis of covariance (ANCOVA), adjusting for baseline measures.Ethics and disseminationEthics approval has been obtained from Unity Health Toronto and The Hospital for Sick Children. Results will be presented locally, nationally and internationally and published in a peer-reviewed journal. The findings may be helpful to nutrition guidelines for children in effort to reduce childhood obesity using a simple, inexpensive and scalable cow’s milk fat intervention.Trial registration numberNCT03914807; pre-results.