The most common reasons for revision of unicompartmental knee arthroplasty (UKA) are loosening and pain. Cementless components may reduce the revision rate. The aim of this study was to compare the fixation and clinical outcome of cementless and cemented Oxford UKAs. A total of 43 patients were randomised to receive either a cemented or a cementless Oxford UKA and were followed for two years with radiostereometric analysis (RSA), radiographs aligned with the bone–implant interfaces and clinical scores. The femoral components migrated significantly during the first year (mean 0.2 mm) but not during the second. There was no significant difference in the extent of migration between cemented and cementless femoral components in either the first or the second year. In the first year the cementless tibial components subsided significantly more than the cemented components (mean 0.28 mm (sd 0.17) vs. 0.09 mm (sd 0.19 mm)). In the second year, although there was a small amount of subsidence (mean 0.05 mm) there was no significant difference (p = 0.92) between cemented and cementless tibial components. There were no femoral radiolucencies. Tibial radiolucencies were narrow (< 1 mm) and were significantly (p = 0.02) less common with cementless (6 of 21) than cemented (13 of 21) components at two years. There were no complete radiolucencies with cementless components, whereas five of 21 (24%) cemented components had complete radiolucencies. The clinical scores at two years were not significantly different (p = 0.20). As second-year migration is predictive of subsequent loosening, and as radiolucency is suggestive of reduced implant–bone contact, these data suggest that fixation of the cementless components is at least as good as, if not better than, that of cemented devices. Cite this article: Bone Joint J 2015; 97-B:185–91.