In a prospective, controlled clinical study we randomised 50 patients with primary coxarthrosis into either removal or retention of the subchondral bone plate during ace-tabular preparation in cemented total hip arthroplasty. The effect was evaluated for a 2-year follow up period by repeated RSA examinations, analyses of radiolucent lines on conventional radiographs and clinical follow-ups with WOMAC, SF-12 and Harris Hip Score. Removal of the subchondral bone plate resulted in an improvement in radiological appearance of the bone-cement interface. For the retention group the extent of radiolucent lines as measured on pelvic and AP-view, had increased from a direct postoperative average level of 3.4% to a 2-year level of 28.8%. For the group with removal of the subchondral bone plate, the direct postoperative radiographs revealed no radiolucency, and at 2 years it only occupied a mean of 4.1 % of the interface. With the classification according to Hodgkinson the retention group had 10 out of 25 patients remaining in grade 0 (no demarcation) at 2years, whereas the removal group had 23 out of 25 patients in grade 0 at 2 years. The RSA results showed small early migration in both groups, but a tendency towards better stability and less scatter of the results in the removal group. The retention group tilted from 6 months onwards slightly but continuously towards a more horizontal position, whereas the removal group stabilized in a slightly vertical position after 1 year. The mean proximal migrations for all cups taken together were 0.09 mm at 2 years with no significant difference between groups. No differences were found in clinical outcome neither pre- nor postoperatively. To optimize the bone-cement interface and thereby increase the long time cup survival, removal of the subchondral bone plate where possible appears to be advantageous, but it is a more demanding surgical technique.