Purpose: While knee joint biomechanics and muscle activation patterns collected during gait analysis provide information about joint loading magnitude and duration during a single step, physical activity data collected with accelerometers provides an objective, surrogate metric of joint loading frequency (step count) during daily life. Together with data from gait analysis, frequency data might provide further insight into the relationships between in vivo joint loading exposure and clinical knee osteoarthritis (OA) progression. To investigate whether joint loading frequency data adds to our understanding of clinical knee OA progression, the following hypotheses were tested on individuals with medial knee OA: 1) joint loading magnitude and pattern metrics (from gait analysis) and joint loading frequency (from accelerometer data) will not be correlated, and 2) individuals who progress clinically - total knee arthroplasty (TKA) endpoint - at 3.5-year follow-up will exhibit baseline differences in gait metrics (matching those previously reported) and lower joint loading frequency compared to those who do not progress. Methods: Fifty-seven individuals with knee OA and primarily medial tibiofemoral compartment involvement underwent standardized three-dimensional instrumented baseline gait analysis during over-ground walking at self-selected speed. Surface electromyography of the quadriceps and hamstrings was collected concurrently. For one week following gait analysis, participants wore a tri-axial accelerometer during waking hours. Principal component analysis extracted major modes of variation (PCs) from knee joint moment waveforms during gait, including PCs describing time varying magnitude and patterns over the gait cycle. Step count was calculated from the accelerometer data using proprietary formulas within ActiLife software.Spearman correlation coefficients were calculated between step count and gait metrics that have been associated with OA progression, including features describing less dynamic patterns in the knee adduction (KAM), flexion (KFM), and rotation (KRM) moments and prolonged quadriceps and hamstrings activation in mid-stance. Jonckheere-Terpstra tests for ordered alternatives examined gait patterns across quartiles of step count with post-hoc comparisons made using Bonferroni corrections. For a subsample of 33 participants, Mann Whitney U-tests examined baseline differences in gait and accelerometer-derived metrics between individuals who progressed (TKA) at 3.5-year follow-up and those who did not. Results: Contrary to our hypothesis, significant correlations were found between step count and gait variables that have been associated with osteoarthritis progression. Specifically, lower step count was associated with a lack of KAM unloading, prolonged activation of the quadriceps and hamstrings in mid-stance, and a smaller range of KFM and KRM. Inspection of gait variables across quartiles of step count revealed that individuals in the lowest quartile (averaging just under 4000 steps/day) exhibited gait patterns that were significantly different than the other quartiles, including smaller KFM range and prolonged quadriceps and hamstrings activity through mid-stance (Table 1, Figure 1). The group that progressed clinically at 3.5-year follow-up had a higher overall KAM magnitude and prolonged vastus medialis activation through mid-stance at baseline compared to those who did not progress, but there were no significant differences in step count at baseline between those who did or did not progress at follow-up. Conclusions: The moment and EMG features that were correlated with low step count and were observed in the lowest quartile of step count have previously been associated with clinical knee OA progression (TKA endpoint) at 8-year follow-up. This may indicate these individuals are modulating both their gait and physical activity levels in response to their OA. Interestingly, there were few differences in gait patterns among those in the highest three quartiles of step count, suggesting there may be a threshold below which individuals exhibit this combination of “at risk” gait patterns and low physical activity levels. It is also of note that many of the gait features associated with step count were describing patterns rather than magnitude of joint loading. While the results of the longitudinal analysis support prior work showing baseline gait differences related to future clinical OA progression, a longer follow-up period may be needed to understand whether this combination of “at risk” gait patterns and low joint loading frequency puts individuals at higher or lower risk for longer-term OA progression.