Purpose Knee adduction moment (KAM) has been recognized as a good clinical surrogate for medial tibiofemoral joint loading and is associated with implant durability after total knee arthroplasty (TKA). This study aimed to examine the effects of joint line obliquity in kinematically aligned TKA (KA-TKA) on KAM during gait. Methods The study enrolled 21 knees from 18 patients who underwent cylindrical axis reference KA-TKA and a matched group of 21 knees from 18 patients who underwent mechanically aligned (MA)-TKA as controls. Gait analyses were performed the day before TKA and at an overall mean of 2.6 years postoperatively. First peak KAM and variables associated with frontal knee kinetics were determined and compared between groups. Results In KA-TKA, the proximal tibia was resected with 3.4° ± 1.5° of varus in relation to the mechanical axis, and the final femorotibial shaft axis was 176.7° ± 3.8° with KA-TKA and 174.4° ± 3.0° with MA-TKA. KAM was significantly smaller with KA-TKA than with MA-TKA ( p  < 0.032). Regarding variables affecting KAM, significant differences were evident between the two TKAs for knee adduction angle ( p  = 0.0021), lever arm ( p  = 0.028), and Δlever arm ( p  = 0.0001). Conclusions In KA-TKA, joint line obliquity reduced peak KAM during gait, despite slight varus limb alignment, and this reduced KAM in KA-TKA can tolerate constitutional varus alignment. In clinical settings, KA-TKA thus represents a promising technical option for patients with large coronal bowing of the shaft carrying a risk of increased KAM after TKA. Level of evidence III.