s u m m a r yObjective: To investigate the longitudinal association between objectively measured ambulatory activity (AA) and knee MRI-detected osteophytes (OPs), and to test whether this relationship was modified by common risk factors for OA including sex, obesity, disease severity and knee injury history. Methods: 408 community-dwelling adults aged 51e81 years were assessed at baseline and 2.7 years. T1weighted fat-suppressed MRI was used to evaluate knee OPs at both time points. AA was assessed at baseline by pedometers and categorized as: less active ( 7499 steps per day), moderately active (7500 e9999 steps per day) and highly active (!10,000 steps per day). Results: Statistically significant interactions were detected between knee OA risk factors and AA on increases in MRI-detected OPs (all P < 0.05). In stratified analyses, being moderately active, compared to being less active, was protective against an increase in MRI-detected OPs (score change of !1) in females (relative risk (RR) ¼ 0.42, 95%CI, 0.25e0.70, P < 0.01), those who were obese (RR ¼ 0.50, 95%CI, 0.30 e0.83, P < 0.01), those with radiographic OA (ROA) (RR ¼ 0.68, 95%CI, 0.47e0.97, P ¼ 0.02) and those with a history of knee injury (RR ¼ 0.27, 95%CI, 0.08e0.88, P ¼ 0.02) in almost every knee compartment, after adjustment for confounders. No statistically significant associations were found in males, non-obese, non-ROA or non-injury groups. Conclusions: Being moderately active is protective against an increase in MRI-detected OPs in females, those with ROA, those who are obese and those with a history of knee injury. These findings suggest that being moderately active is beneficial for individuals who are at higher risk of knee OA.