41 consecutive knee arthroplasties were randomized to either computer-navigated or non-navigated to investigate whether navigation improved soft-tissue balancing in the procedure. Mediolateral flexion and extension gaps were measured to determine the accuracy of balancing. The accuracy of bony cuts and limb alignment were also assessed. The results indicated that navigation was associated with more accurate extension gap balancing. Despite this, there was no significant difference in flexion gap balancing or in obtaining equal flexion and extension gaps.