We compared lower limb coronal alignment measurements obtained pre- and post-operatively with long-leg radiographs and computer navigation in patients undergoing primary total knee replacement (TKR). A series of 185 patients had their pre- and post-implant radiological and computer-navigation system measurements of coronal alignment compared using the Bland-Altman method. The study included 81 men and 104 women with a mean age of 68.5 years (32 to 87) and a mean body mass index of 31.7 kg/m(2) (19 to 49). Pre-implant Bland-Altman limits of agreement were -9.4° to 8.6° with a repeatability coefficient of 9.0°. The Bland-Altman plot showed a tendency for the radiological measurement to indicate a higher level of pre-operative deformity than the corresponding navigation measurement. Post-implant limits of agreement were -5.0° to 5.4° with a repeatability coefficient of 5.2°. The tendency for valgus knees to have greater deformity on the radiograph was still seen, but was weaker for varus knees. The alignment seen or measured intra-operatively during TKR is not necessarily the same as the deformity seen on a standing long-leg radiograph either pre- or post-operatively. Further investigation into the effect of weight-bearing and surgical exposure of the joint on the mechanical femorotibial angle is required to enable the most appropriate intra-operative alignment to be selected.