INTRODUCTION & AIMS Painful exercise reduces pain in the short-term compared to non-painful exercise in people with chronic pain. However, integration of painful exercise into knee osteoarthritis (OA) treatment is contentious among clinicians, with the clinical implications unknown. This study aimed to explore the feasibility of painful versus non-painful exercise in people with knee OA. METHODS Participants were randomised into a painful (INT; RPE 5-6) or non-painful (CON; RPE 4-5) who performed exercise twice per week for 6 weeks. Both groups also received standardised education. The primary outcomes were feasibility, assessed using participant recruitment, retention, adherence, and compliance rates. Secondary outcomes included pain, function, strength, pressure pain thresholds, psychosocial measures and adverse events. Immediate pre- and post-exercise pain were also measured. RESULTS 21 people with knee OA (66 ± 9 years old; BMI: 29.4 ± 8.1) completed the study. Feasibility was achieved and supported by high rates of recruitment (INT = 89%, CON 89%), retention (INT = 91%, CON = 100%), adherence (INT = 91%, CON = 92%), and compliance (INT = 72%, CON = 81%). Pain reduced in both groups (mean difference [95% CI]; INT = -0.7 [-1.8 to 0.4]; CON = -1.5 [-2.7 to 0.4]), and strength (1-RM leg press) improved with a moderate effect in favour of INT (12.8 [0.2 to 25.9], d = 0.97, p = 0.046). CONCLUSION Incorporating painful exercise into treatment for individuals with knee OA is feasible. Several potential benefits include positive changes in maladaptive beliefs and behaviours and enhanced systemic benefits associated with higher intensity exercise. Future research comparing the efficacy of painful versus non-painful exercise in knee OA and other chronic musculoskeletal conditions is warranted.