Aim <p>This study aimed to ascertain whether a discrepancy exists between coronal deformity measurements obtained from radiographs and those from computer navigation, and to explore potential reasons for such discrepancy.</p> Methods <p>This study assessed 621 primary navigated TKAs performed on varus arthritic knees. Preoperative and postoperative hip–knee–ankle angle (HKA), mechanical lateral distal femoral angle (LDFA), medial proximal tibial angle (MPTA), and joint congruence angle (JCA) were measured using scanograms. Initial coronal deformity after arthrotomy and prior to performing releases or bone cuts was recorded during navigation using a proven dynamic method. The difference between radiographic and navigation measurements of initial deformity was calculated.</p> Results <p>The mean preoperative HKA was 12.6º (SD ± 6.11, range 0.8º to 40.9º), while the mean initial alignment on navigation was 14° (SD ± 5.96, range 2º to 42º). The mean difference between these values of—1.42º was highly significant (SD ± 3.04, range -17º to 10.5º, <i>P</i> &lt; 0.001 on <i>t </i>test,), with a correlation of r = 0.873 (<i>P</i> &lt; 0.001). This difference exceeded ± 2º in over half of the knees (<i>n</i> = 314, 50.6%). There was a progressively increasing difference between the two measurements with an increase in mean initial HKA, LDFA, and JCA, and a decrease in mean MPTA (ANOVA, <i>P</i> &lt; 0.001).</p> Conclusion <p>Surgeons performing TKA surgery using navigation or robots should be aware of the potential discrepancies between coronal deformity measurements. Factors such as increased coronal deformity, medial bone deficiency, and lateral laxity may contribute to this discrepancy.</p>