Although bone cement is the primary mode of fixation in total knee arthroplasty (TKA), cementless fixation is gaining interest as it has the potential of achieving lasting biological fixation. By 3D printing an implant, highly porous structures can be manufactured, promoting osseointegration into the implant to prevent aseptic loosening. This study compares the migration of cementless, 3D-printed TKA to cemented TKA of a similar design up to two years of follow-up using radiostereometric analysis (RSA) known for its ability to predict aseptic loosening. Methods A total of 72 patients were randomized to either cementless 3D-printed or a cemented cruciate retaining TKA. RSA and clinical scores were evaluated at baseline and postoperatively at three, 12, and 24 months. A mixed model was used to analyze the repeated measurements. Results The mean maximum total point motion (MTPM) at three, 12, and 24 months was 0.33 mm (95% confidence interval (CI) 0.25 to 0.42), 0.42 mm (95% CI 0.33 to 0.51), and 0.47 mm (95% CI 0.38 to 0.57) respectively in the cemented group, versus 0.52 mm (95% CI 0.43 to 0.63), 0.62 mm (95% CI 0.52 to 0.73), and 0.64 mm (95% CI 0.53 to 0.75) in the cementless group (p = 0.003). However, using three months as baseline, no difference in mean migration between groups was found (p = 0.497). Three implants in the cemented group showed a > 0.2 mm increase in MTPM between one and two years of follow-up. In the cementless group, one implant was revised due to pain and progressive migration, and one patient had a liner-exchange due to a deep infection. Conclusion The cementless TKA migrated more than the cemented TKA in the first two-year period. This difference was mainly due to a higher initial migration of the cementless TKA in the first three postoperative months after which stabilization was observed in all but one malaligned and early revised TKA. Whether the biological fixation of the cementless implants will result in an increased long-term survivorship requires a longer follow-up.@en