Background. Adequate rotation of femoral component in total knee arthroplasty(TKR) is mandatory for preventing numerous adverse sequelae. The transepicondylar axis has been a well-accepted reference for femoral component rotation in the measured resection technique. In this technique, measured resection is performed referenced off the tibial cut - perpendicular to the tibial mechanical axis with the knee in 90 ° of flexion. However, to the best of our knowledge, it is not known whether this technique apply well to a knee with tibia vara. This study evaluates the reliability of the transepicondylar axis as a rotational landmark in knees with tibia vara. Methods. We selected 101 osteoarthritis knees in 84 symptomatic patients(mean age: 69.24 ± 5.68) with proximal tibia vara (Group A). Group A was compared with 150 osteoarthritic knees without tibia vara in 122 symptomatic patients (mean age: 69.51 ± 6.01) (Group B). The guide line for selection of all these knees were based on the degree of tibia vara angle (TVA) which was formed by line perpendicular to epiphysis and by anatomical axis of the tibia - all measured in radiographs of the entire lower limb. Magnetic resonance imaging (MRI) axial images with most prominent part of both femoral condyles were used for measurement of transepicondylar axis(TE), anteroposterior axis(AP) and posterior condylar axis(PC). Results. The mean TVA of group A was 8.94° ± 3.11 and group B was 1.24° ± 0.85. The TE line in Group A showed 6.09 ° ± 1.43 of external rotation, relative to PC. This did not show statistical difference compared with 5.95 ° ± 1.58 in Group B (p=0.4717). The AP line in Group A showed 6.06 ° ± 1.93 of external rotation, relative to the line perpendicular to PC. This was statistically significant when compared to 5.44 ° ± 2.13 in Group B (p=0.020). Conclusion. There is no difference between knees without tibia vara compared those with tibia vara with regards to transepicondylar axis. In addition, both groups have almost identical external rotation of approximately 6 °. The AP axis was only approximately 0.5 ° difference between the two groups. The distal femoral geometry was not affected by tibia vara deformity, that is, there were no hypoplastic or hyperplastic deformities of medial femoral condyle in osteoarthritic knees with tibia vara. The use of transepicondylar axes in determining femoral rotation may produce flexion asymmetry in knees with proximal tibia vara. So, It should be pointed out that more attention should be paid on femoral component rotation and flexion gap balancing in knees with proximal tibial vara