Background and purpose — Bone remodeling as a response to bone trauma, postoperative immobilization, and device-related bone reactions can lead to loss of bone stock and increase the risk of periprosthetic fracture and aseptic loosening. This study investigates the adaptive bone remodeling of the proximal tibia after uncemented total knee arthroplasty (TKA). Patients and methods — We performed a 2-year follow up of 53 patients (mean age 62 (38–70) years, 27 of whom were men, who received an uncemented TKA in a randomized controlled trial with bone mineral density (BMD) as secondary endpoint. Patients were randomized to 2 groups of either monoblock (A) or modular (B) polyethylene design. The TKAs were performed using the uncemented Zimmer Nexgen trabecular metal. Measurements of BMD were done postoperatively and after 3, 6, 12, and 24 months. BMD was measured in 3 regions of interest (ROI). Results and interpretation — In group A statistically significant changes in BMD were seen after 24 months in both the medial and lateral ROI. BMD decreased medially by 15% (p = 0.004) and laterally by 13% (p = 0.01). In group B the BMD changes were limited and after 24 months returned to the preoperative values. The differences in BMD change between groups were statistically significant in both the medial (p = 0.03) and lateral (p = 0.02) ROI. In the distal ROI we found no significant change in BMD in either group. A significantly different bone remodeling pattern of the proximal tibia was seen in the 2 groups with a higher degree of bone loss in the knees that received the monoblock polyethylene design, indicating that the flexible monoblock implant design, previously shown to improve fixation, does not decrease the bone loss of the proximal tibia.