Background <p>Knee osteoarthritis (KOA) is the most common and burdensome form of osteoarthritis worldwide. Although numerous non-pharmacological interventions are recommended in clinical practice, existing evidence remains fragmented and lacks comprehensive comparative evaluation. This study aims to systematically compare the effectiveness of first- and second-line non-pharmacological interventions recommended by the American Academy of Orthopaedic Surgeons (AAOS) for symptomatic KOA using Bayesian network meta-analysis (NMA), which integrates both direct and indirect evidence from randomized controlled trials (RCTs).</p> Methods <p>We will conduct a systematic review and Bayesian NMA of randomized trials evaluating non-pharmacological interventions for symptomatic KOA. Eligible interventions include exercise, weight management, braces, and other guideline-recommended non-pharmacological therapies with strong or moderate evidence. The primary outcome is pain intensity, assessed using validated scales. Secondary outcomes include physical function, quality of life, stiffness, global assessment, and adverse events. A comprehensive search will be conducted across multiple databases without language restrictions from inception to the date of the formal search. Risk of bias will be assessed using the Cochrane RoB 2.0 tool. Bayesian NMA models will be implemented using Markov chain Monte Carlo methods. Subgroup analyses and meta-regressions will explore heterogeneity related to disease severity and treatment characteristics. Certainty of evidence will be assessed using the GRADE approach adapted for NMA.</p> Conclusions <p>This NMA will synthesize high-quality evidence on the comparative effectiveness of AAOS-recommended non-pharmacological interventions for KOA. The findings will provide clinically relevant treatment rankings, support individualized non-pharmacological management in adults, and inform future guideline updates and research priorities.</p> Systematic registration number <p>PROSPERO CRD420251045765.</p>