Background: The component design and fixation method of joint arthroplasty may affect component migration and survival. The aim of this study was to compare fixation of cementless twin-peg (CLTP), cemented twin-peg (CTP) and cemented single-peg (CSP) femoral components of medial unicompartmental knee replacement (UKR). Methods: Eighty patients (mean age = 63 years, 48 males) with medial knee osteoarthritis were randomized in three ways to CLTP (n = 25), CTP (n = 26) or CSP (n = 29) femoral UKR components. The patients were followed 5 years postoperatively with RSA, bone mineral density (BMD), PROMs and radiological evaluation of radiolucent lines (RLL), femoral component flexion angle and complications. Results: At the 5-year follow-up, femoral component total translation was comparable between the three groups (p = 0.60). Femoral component internal rotation was 0.50° (95% CI 0.3; 0.69) for the CLTP group, 0.58° (95% CI 0.38; 0.77) for the CTP group and 0.25° (95% CI 0.07; 0.43) for the CSP group (p = 0.01). BMD decreased peri-prosthetically (range − 11.5%; − 14.0%) until 6-month follow-up and increased toward the 5-year follow-up (range − 3.6%; − 5.8%). BMD change did not correlate with component migration. Lower flexion angle was correlated with higher 5-year subsidence, total translation, varus rotation and maximum total point motion (p = 0.01). Two patients (1 CLTP, 1 CTP) had RLL in the posterior zone. There were two revisions. Conclusion: At 5-year follow-up, fixation of UKA femoral components with twin-peg was not superior to the single-peg design. Cementless and cemented twin-peg femoral components had similar fixation. A lower flexion angle was correlated with higher component migration.