Abstract Objective Investigate the association between physical activity and pain severity in individuals with knee osteoarthritis. Design Cross-sectional; systematic review with meta-analyses. Methods Thirty-one participants with knee osteoarthritis underwent assessment of symptoms via self-report questionnaires and quantitative sensory testing. Following testing, physical activity and symptoms were monitored for seven days using accelerometers and logbooks. Cross-correlation analyses were performed on fluctuations in symptoms and physical activity across the week to detect the relative timing of the strongest association between pain and activity. These data were complemented by meta-analyses of studies that examined correlations between pain from knee osteoarthritis and physical activity or fitness. Results Pain severity at baseline correlated with moderate to vigorous physical activity (r2 = 0.161–0.212, P < 0.05), whereby participants who were more physically active had less pain. Conversely, the peak of the cross-correlation analyses was most often positive and lagging, which indicated that pain was increased subsequent to periods of increased activity. These superficially discrepant findings were supported by the results of a meta-analysis of 13 studies and 9,363 participants, which identified significant heterogeneity for associations between physical activity and pain (I2 = 91%). Stronger inverse associations were found between fitness and pain. Conclusions Associations between physical activity and pain in people with knee osteoarthritis are variable and dynamic. These results reflect the beneficial impact of an active lifestyle and accompanying higher fitness. Yet, the side effect of acute periods of physical activity to transiently exacerbate pain may influence the behavior of some people to avoid activity because of pain.