ObjectiveOur primary aim is to determine the effect of a six-week toe-in, toe-out and active placebo gait retraining program on proxy measures of medial knee joint load and varus thrust in people with medial knee osteoarthritis. Our secondary aim is to determine the intervention effects on patient reported outcomes and physical function and determine if changes are maintained at three-months follow-up. MethodsWe will conduct a three-arm randomised placebo-controlled trial. Ninety participants with medial knee osteoarthritis will be randomised and stratified via varus thrust status (presence/absence) to: toe-in, toe-out or placebo gait retraining (an intervention that does not change proxy measures of medial knee joint load). The intervention involves weekly clinician-supervised sessions with biofeedback, knee osteoarthritis education, motor learning and behaviour change principles, and daily gait retraining practice. Primary outcomes are proxy measures of medial knee joint load: knee adduction moment (early- and late-stance peaks and impulse), and varus thrust (presence/absence). Secondary outcomes include pain, physical function, medication and health care utilisation, quality of life, work ability, treatment blinding, intervention credibility and other biomechanical outcomes. Assessment timepoints are at baseline, six weeks (post intensive training), and three-months following the six-week intervention. ConclusionOur trial will determine whether toe-in or toe-out gait retraining is most effective at reducing proxy measures of medial knee joint load and varus thrust in people with medial knee osteoarthritis. This study will also evaluate if toe-in or toe-out gait retraining interventions are superior at improving pain, physical function and quality of life compared to placebo. Clinical trial registrationThis clinical trial protocol is registered with the Australian New Zealand Clinical Trials Registry (ACTRN12621000414819).