The role of physical activity (PA) in knee osteoarthritis (OA) remains unclear. Some studies found engaging in moderate-to-vigorous PA (MVPA) had no effect on knee OA, while others found MVPA was harmful. This lack of agreement is explained by the use of traditional regression models, which fail to consider that increasing time in MVPA also decreases time spent in other intensities, such as light PA or sedentary time, since the amount of time in a day is fixed. This interplay can be addressed by an advanced technique called compositional data analysis (CoDA). Thus, we used CoDA to examine how increasing time in one intensity (MVPA, light PA, or sedentary time), while reducing time spent in the other two intensities, affected knee OA progression over 2 years. We used data from a large cohort study called the Osteoarthritis Initiative. Participants wore a research-grade activity monitor that counted minutes spent in each intensity of PA (MVPA, light PA, sedentary time). Participants also had knee X-rays at baseline and 2-year follow-up, which were used to measure knee OA progression. We used three separate CoDA models, one for each intensity, to estimate how increasing time in that intensity, while decreasing time in the other intensities, affected knee OA progression (adjusting for potential confounders). We included 969 OAI participants. On average, participants spent 18 minutes in MVPA, 284 in light PA, and 677 in sedentary time. We found that increasing MVPA resulted in less OA progression over 2 years. Interestingly, increasing sedentary time also resulted in less OA progression, while increasing light PA resulted in more OA progression. Unlike previous studies, we used CoDA and found that increasing MVPA (while reducing light PA and sedentary) resulted in less OA progression. This means that engaging in MVPA may actually be beneficial for adults with knee OA.