Objectives To firstly investigate the efficacy of three different dosages of home-based, knee-extensor strength exercise on isometric knee-extensor strength in patients eligible for knee replacement due to severe knee osteoarthritis, and secondly, the influence of exercise on symptoms, physical function and decision on surgery. Design Randomized dose-response trial. Setting Primary and secondary care. Participants One-hundred and forty patients eligible for knee replacement were included. Eligibility for surgery was assessed by an orthopedic surgeon specialized in knee arthroplasty. Interventions Patients were randomized to three groups; 2, 4 or 6 home-based knee-extensor exercise-sessions per week (group 2, 4 and 6 respectively) for 12 weeks (N=47/group). Exercise instruction was given by a trained physiotherapist. Main outcome measures The primary outcome was isometric knee-extensor strength. Secondary outcomes were Oxford Knee Score, Knee Osteoarthritis Outcome Score, average knee pain last week (0-10 numeric rating scale), 6-minute walk test, stair climbing test, exercise adherence and 'need for surgery'. The primary endpoint was after 12 weeks of exercise (before surgery) and the secondary after surgery. Outcome assessors and patients were blinded to allocation. Results After 12 weeks of exercise, data were available for 117 patients (N=39/group). Isometric knee-extensor strength increased in all groups but intention-to-treat analysis showed no difference between the three groups in change from baseline to after 12 weeks of exercise in isometric knee-extensor muscle strength: Group 2 vs. 4 (0.009 Nm/kg [95% CI -0.15 to 0.17], P=0.913), group 2 vs. 6 (-0.03 Nm/kg [95% CI -0.18 to 0.13], P=0.725) and group 4 vs. 6 (-0.04 Nm/kg [95% CI -0.20 to 0.13], P=0.668). For the secondary outcomes a significant difference before surgery was found between group 2 and group 6 for Oxford Knee Score (4.2 [95% CI 0.6 to 7.8], P=0.02) and average knee pain last week (NRS 0-10) (-1.1 [95% -2.2 to -0.1], P=0.03) in favour of group 2. No other differences were observed before and after surgery. After the 12-week exercise intervention, 38 (32.5%) patients wanted surgery and 79 (67.5%) postponed surgery. This was independent of exercise dosage. Conclusion Knee-extensor strength increased with both 2, 4 and 6 exercise sessions per week but none of the prescribed exercise dosages were superior to the others for the primary outcome isometric knee-extensor strength after 12 weeks. Two home-based knee-extensor exercise sessions a week seems superior to six for patient-reported outcomes, and across exercise dosages, only one in three patients decided to have surgery after the coordinated home-based exercise intervention. Trial registration ClinicalTrials.gov identifier: NCT02931058, pre-registered October 10th, 2016. Protocol PubMed (PMID: 29347947).