The gold standard for measuring knee alignment is mechanical axis determined using full-limb radiographs (FLR). Measurement of joint alignment using antero-posterior (AP) knee radiographs is more accessible, economical and involves less radiation exposure to the patient compared with using full-limb radiographs. The aim of this study was to compare and assess the reproducibility of knee joint axial alignment on full-limb radiographs and conventional AP knee radiographs. Knee alignment was measured in 40 subjects (80 knees) from the TwinsUK registry. Measurement of mechanical knee alignment was from FLR, and anatomic knee alignment from weight-bearing AP knee radiographs. Reproducibility was assessed by intra-class correlation coefficients and kappa statistics. Reproducibility of knee alignment for both methods was good, with intra-observer ICC's of 0.99 for both FLR and AP radiographs. The mean alignment angle on FLR was 178.9 degrees (SD 2.1, range 173-183 degrees ), and 179.0 degrees (SD 2.1, range 173-185 degrees ) on AP films. 58.8% of knees on FLR and 66.3% on AP films were of varus alignment. Good correlations were seen between results for FLR and AP radiographs, with ICC ranging from 0.87-0.92 for left and right knees, and kappa statistics of 0.65-0.74. Standard AP knee radiographs can be used to measure knee alignment with good reproducibility, and provide comparable results to those obtained from FLR. This will facilitate measurement of knee alignment in existing cohort studies to assess malalignment as a risk factor of incident OA, and in clinical practice.