We evaluated a behaviorally designed intervention utilizing gamification and social support to improve physical activity and reduce symptoms in patients with osteoarthritis of the knee (KOA). Veterans with KOA, aged 40-80 years, were enrolled in this randomized controlled trial. Participants received a Fitbit and completed a 2- to 4-week baseline period. A Web-based platform administered biweekly surveys after randomization and tracked physical activity. Participants selected a daily step goal that was 33%, 40%, or 50% above their baseline. The intervention arm received game playing aspects and a social support partner to advance weekly step performance while the control arm only received weekly updates. The primary outcome was the change in steps per day averaged over 2-week intervals. We used mixed effects regression, adjusting for baseline step count. Secondary outcomes assessed the change in KOOS (Knee Injury and Osteoarthritis Outcome Score) over 32 weeks. Thirty-one participants were included in the final analysis. Most participants were male (90.3%), Black (70.96%), had a mean (SD) age of 60 (13) years, and body mass index of 33.7 (5.9) kg/m2. Participants that received the intervention walked a total of 1119 (95% confidence interval: -562, 2799) more steps per day (p = 0.19). The effect was greatest in the first 6 months (1491 [-272, 3254], p = 0.10). Compared with controls, those that received the intervention had improvement over time in total KOOS (mean 2-week change +0.62 [0.031, 1.20] vs -0.38 [-1.04, 0.28], p = 0.02) and several subscales. This intervention demonstrated promise for promoting greater physical activity and improving symptoms in patients with KOA.