Aims. The transepicondylar axis is a well-established reference for the determination of femoral component rotation in total knee arthroplasty (TKA). However, when severe bone loss is present in the femoral condyles, rotational alignment can be more complicated. There is a lack of validated landmarks in the supracondylar region of the distal femur. Therefore, the aim of this study was to analyze the correlation between the surgical transepicondylar axis (sTEA) and the suggested dorsal cortex line (DCL) in the coronal plane and the inter- and intraobserver reliability of its CT scan measurement. Methods. A total of 75 randomly selected CT scans were measured by three experienced surgeons independently. The DCL was defined in the coronal plane as a tangent to the dorsal femoral cortex located 75 mm above the joint line in the frontal plane. The difference between sTEA and DCL was calculated. Descriptive statistics and angulation correlations were generated for the sTEA and DCL, as well as for the distribution of measurement error for intra- and inter-rater reliability. Results. The external rotation of the DCL to the sTEA was a mean of 9.47° (SD 3.06°), and a median of 9.2° (IQR 7.45° to 11.60°), with a minimum value of 1.7° and maximum of 16.3°. The measurements of the DCL demonstrated very good to excellent test-retest and inter-rater reliability coefficients (intraclass correlation coefficient 0.80 to 0.99). Conclusion. This study reveals a correlation between the sTEA and the DCL. Overall, 10° of external rotation of the dorsal femoral cortical bone to the sTEA may serve as a reliable landmark for initial position of the femoral component. Surgeons should be aware that there are outliers in this study in up to 17% of the measurements, which potentially could result in deviations of femoral component rotation. Cite this article: Bone Jt Open 2024;5(12):1067–1071