The use of a computer navigation system is intended to optimize implant positioning. This study compares prospectively followed total knee arthroplasty (TKA) done with imageless computer-assisted navigation with a cohort of TKA done with manual instruments. Primary TKAs were performed on 147 knees, with 78 in the manual group and 69 in the navigated group. Coronal alignment was determined by postoperative full-length standing radiographs. The 2 groups had similar mean values for radiographic mechanical axis at 0.4 degrees . There was a larger variation in alignment in the manual group, with 58% of the manual group within 2 degrees of neutral compared with 78% of the navigated group (P = .008). The navigation system reproducibly resulted in a more reliable reproduction of the mechanical axis.