Background and purpose - The stem on the tibial component of total knee arthroplasty provides mechanical resistance to lift-off, shear forces, and torque. We compared tibial components with finned stems (FS) and I-beam block stems (IS) to assess differences in implant migration. Patients and methods - In a patient-blinded RCT, 54 patients/knees (15 men) with knee osteoarthritis at a mean age of 77 years (70-90) were randomly allocated to receive tibial components with either a FS (n = 27) or an IS (n = 27). Through 5 to 7 years' follow-up, implant migration was measured with RSA, periprosthetic bone mineral density (BMD) was measured with DXA, and surgeons reported American Knee Society Score (AKSS). Results - At minimum 5 years' follow-up, maximum total point motion (MTPM) was higher (p = 0.04) for IS (1.48 mm, 95% CI 0.81-2.16) than for FS (0.85 mm, CI 0.38-1.32) tibial components. Likewise, total rotation (TR) was higher (p = 0.03) for IS (1.51˚, CI 0.78-2.24) than for FS (0.81˚, CI 0.36-1.27). Tibial components with IS externally rotated 0.50° (CI -0.06 to 1.06) while FS internally rotated 0.09° (CI -0.20 to 0.38) (p = 0.03). Periprosthetic bone stress-shielding was higher (p < 0.01) up to 2 years' follow-up for IS compared with FS in the regions medial to the stem (-13% vs. -2%) and posterior to the stem (-13% vs. -2%). Below the stem bone loss was also higher (p = 0.01) for IS compared with FS (-6% vs. +1%) up to 1-year follow-up. Knee score improved similarly in both groups up to 5 years' follow-up. Interpretation - Periprosthetic bone stress-shielding medial and posterior to the stem until 2 years, and tibial component migration at 5 years, was less for a finned compared with an I-shaped block stem design.