Knee osteoarthritis is a common joint problem leading to an increase of pain and a loss of function in older individuals. The main objective of this study was to evaluate if a participant who was randomly assigned to his preferred group improved his adherence to an effective walking program compared to a participant who did not receive his preferred group. This was a 9-month pilot randomized clinical trial, based on a patient treatment preferences design. The 69 eligible participants had a diagnosis of knee osteoarthritis. Participants were randomized to one of two groups: a supervised community-based or unsupervised walking program, based on the Ottawa Panel guidelines. At 6 months, participants who expressed a preference, either for the supervised or unsupervised program, and who were assigned to their preferred choice of program showed significantly higher adherence to walking sessions (supervised 60.7 ± 12.3%, P < 0.0001; unsupervised 43.1 ± 12.1%, P = 0.03), compared to the participants who did not obtain their preferred choice of program. After 9 months, significant improvements were shown according to the level of stiffness evaluated with the WOMAC (P = 0.01) and the functional status assessed with the Timed Up and GO Test (P = 0.04), among the adherent participants who obtained their preference, as compared to those who did not receive their preference. We show this approach promotes long-term adherence to a community-based walking program, while ensuring the maintenance of clinical benefits of walking, among older adults susceptible to avoid or not properly engage in physical activity.