The modern era of endoprosthetic joint replacement started with the introduction of acrylic cement to improve component fixation. Long-term results have, however, indicated that prosthetic fixation remains critical; loosening at the bone-cement interface has become an important problem. Research in recent years has focused on attempts to achieve better fixation by improving cementing techniques, improving prosthetic design by, for example, adding metal support of polyethylene components and by exploring alternative ways to bond prosthetic components to bone without cement. The mechanical integrity of the bone-cement interface has been studied under laboratory conditions. Because of the in-vivo reaction of bone, with the interposition of a fibrous tissue layer at the interface, such studies are not totally valid. Studies on autopsy material, more closely resembling the in-vivo situation, are few and there has been only one previous study like the present one. In this study, roentgen stereophotogrammetric analysis (RSA) was evaluated and found to have an accuracy ten times better than conventional radiography. This accuracy was judged adequate for studies of micromotion. In this work, two types of micromotion of the tibial component were studied; migration, i.e. gradual motion over time, and inducible displacement, i.e. instant motion in response to external forces. Ninety-six knee arthroplasties for gonarthrosis, representing four different types of fixation were studied by roentgen stereophotogrammetric analysis (RSA). Eighty-nine arthroplasties were clinically successful. The follow-up ranged from two to five years. Full post-operative weight-bearing was allowed for all patients, except those operated with a Freeman-Samuelson prosthesis, who were adviced to use crutches for six weeks and partial weight-bearing for another six weeks. Fifty-one conventionally cemented all-polyethylene prostheses, 27 total and 24 unicompartmental, represented a baseline series. Migration was found for all prostheses, with a mean maximum deflection of 1.2 and 0.9 mm, respectively, after four years. In both groups, the major part of the migration occurred during the first year, after which the majority of the components did not migrate further. Some prostheses, with larger migration during the first year, continued to migrate throughout the investigation. None of the total, but the majority of the unicompartmental prostheses showed signs of cold flow within the polyethylene. All prostheses showed reversible inducible displacement, the maximum deflection ranging from 0.2 to 1.0 mm.(ABSTRACT TRUNCATED AT 400 WORDS)