OBJECTIVES: To establish the relative effectiveness and cost of providing a home-based exercise programme versus home-based exercise supplemented with an 8-week class-based exercise programme. DESIGN: The trial was a pragmatic, single-blind randomised clinical trial accompanied by a full economic evaluation. SETTING: Patients were randomly allocated to either home-based exercise or home exercise supplemented with class exercise programmes. PARTICIPANTS: A total of 214 patients, meeting the American College of Rheumatology's classification of knee osteoarthritis, were selected from referrals from the primary and secondary care settings. INTERVENTIONS: Both groups were given a home exercise programme aimed at increasing lower limb strength, and endurance, and improving balance. The supplemented group also attended 8 weeks of twice-weekly knee classes run by a physiotherapist. Classes represented typical knee class provision in the UK. MAIN OUTCOME MEASURES: Assessments of locomotor function, using a timed score of three locomotor activities, walking pain and self-reported disability with the Western Ontario and McMaster's Universities osteoarthritis index (WOMAC) were made. General health, lower limb strength, range of movement and compliance with exercise were also measured. Patients were assessed before and after treatment, and also at 6- and 12-month follow-ups. The economic evaluation looked at health service resource use and assessed cost-effectiveness by relating differential costs to differences in quality-adjusted life-years (QALYs) based on patients' responses to the EuroQol-5 Dimensions. Data were obtained at baseline, 1 month, 6 months and 12 months through face-to-face interviews and, where appropriate, examination of hospital medical records. RESULTS: Patients from the supplemented group demonstrated significantly greater improvement in locomotor function and decrease in pain while walking at all follow-ups. The supplemented group also demonstrated smaller but significant improvements in balance, strength, WOMAC score, and the physical function and pain dimensions of the Short Form-36. However, not all of these improvements were maintained over the 12-month follow-up period. There was no evidence that compliance with the home exercise programme was different or that total costs or mean QALY gains were significantly different between the groups. However, costs were slightly lower and QALY gains slightly higher in the group with the supplementary class-based programme. The economic evaluation suggests that supplemented programmes are likely to be considered cost-effective, although there is uncertainty around this estimate, with approximately 30--35% probability that the intervention would not be cost-effective. CONCLUSIONS: The supplementation of a home-based exercise programme with a class-based exercise programme led to superior improvement in the supplemented group. These differential improvements were still evident at review 12 months after treatment had ceased. The additional cost of the supplemented group was offset by reductions in resource use elsewhere in the system. Compliance with the home exercise programme did not differ between the groups. Based on this evidence, the supplementation of a home-based exercise programme with an 8-week class-based exercise programme can be confidently expected to produce small improvements in locomotor function and clinically important reductions in pain. It is recommended that future research investigates methods of increasing compliance with home exercise programmes and evaluates the impact of these interventions in the primary care setting, where most patients with knee osteoarthritis are managed.