Background: While several studies describe kinematics of healthy and osteoarthritic knees using the accurate imaging and computer modelling now possible, no systematic review exists to synthesise these data. Method: A systematic review extracted quantitative observational, quasi-experimental and experimental studies from PubMed, Scopus, Medline and Web of Science that examined motion of the bony or articular surfaces of the tibiofemoral joint during any functional activity. Studies using surface markers, animals, and in-vitro studies were excluded. Results: 352 studies were screened to include 23 studies. Dynamic kinematics were recorded for gait, step-up, kneeling, squat and lunge and quasi-static squat, knee flexion in side-lying or supine leg-press. Kinematics were described using a diverse range of measures including six degrees of freedom kinematics, contact patterns or the projection of the femoral condylar axis above the tibia. Meta-analysis of data was not possible since no three papers recorded the same activity with the same measures. Visual evaluation of data revealed that knees with osteoarthritis maintained a more adducted position and showed less posterior translation of the lateral femoral condylar axis than healthy knees. Variability in activities and in recording measures produced greater variation in kinematics, than did knee osteoarthritis. Conclusion: Differences in kinematics between osteoarthritic and healthy knees were observed, however, these differences were more subtle than expected. The synthesis and progress of this research could be facilitated by a consensus on reference systems for axes and kinematic reporting.