Background: Proponents of minimally invasive total knee arthroplasty argue that retracting rather than everting the patella results in quicker postoperative recovery and improved function. We aimed to investigate this in patients undergoing knee arthroplasty through a standard medial parapatellar approach. Methods: In a prospective randomized double-blinded study, sixty-eight patients undergoing total knee arthroplasty through a standard medial parapatellar approach were assigned to either retraction or eversion of the patella. Postoperatively, at three months, and at one year after surgery, an independent observer assessed the primary outcome measure (i.e., knee flexion) and secondary outcome measures (i.e., Oxford knee score, Short Form-12 [SF-12] score, visual analog scale pain score, knee motion, and alignment and patellar height as measured on radiographs with use of the Insall-Salvati ratio). Results: Early (three-month) follow-up showed no significant difference between patellar eversion and subluxation in flexion (mean and 95% confidence interval [CI], 101° ± 5.37° versus 102° ± 4.14°, respectively), Oxford knee scores (25 ± 3 versus 27 ± 2.69, respectively), SF-12, or visual analog scale pain scores (1.9 ± 0.54 versus 1.1 ± 0.44, respectively). A significant improvement in extension was found (23.9° ± 1.12° versus 22.0° ± 0.91°, respectively [p = 0.034]), but this was not clinically significant. There was no significant difference in any of the outcomes at one year. There was a significant difference in implant malpositioning between the eversion group and the subluxation group, with an increased percentage of lateral tibial overhang in the subluxation group (0.45 ± 0.39 versus 1.84 ± 0.82, respectively [p = 0.005]), but this did not correlate with functional outcome. There was no significant difference in alignment between the two groups (178.29° ± 0.84° versus 178.18° ± 0.78°). At one year after surgery, there was no difference between the two groups in Insall-Salvati ratio (1.15 ± 0.06 versus 1.12 ± 0.06) although there was a correlation between the percentage reduction in the ratio and functional outcome. There were two partial divisions of the patella tendon in the subluxation group, but no patella-related complications in the eversion group. Conclusions: The results of this trial showed that retracting rather than everting the patella during total knee arthroplasty resulted in no significant clinical benefit in the early to medium term. We observed no increase in patellar tendon shortening as a result of eversion rather than subluxation. Our findings did suggest that, with subluxation, there may be an increased risk of damage to the patellar tendon and reduced visualization of the lateral compartment, leading to an increase in implant malpositioning with lateral tibial overhang. Level of Evidence: Therapeutic Level I. See Instructions for Authors for a complete description of levels of evidence. Copyright © 2014 by the journal of bone and joint surgery, incorporated.