Aims The aim of this study was to determine the stability of a new short femoral stem compared with a conventional femoral stem in patients undergoing cementless total hip arthroplasty (THA), in a prospective randomized controlled trial using radiostereometric analysis (RSA). Patients and Methods A total of 53 patients were randomized to receive cementless THA with either a short femoral stem (MiniHip, 26 patients, mean age: 52 years, nine male) or a conventional length femoral stem (MetaFix, 23 patients, mean age: 53 years, 11 male). All patients received the same cementless acetabular component. Two-year follow-up was available on 38 patients. Stability was assessed through migration and dynamically inducible micromotion. Radiographs for RSA were taken postoperatively and at three, six, 12, 18, and 24 months. Results At two years, there was significantly less subsidence (inferior migration) of the short femoral stem (head, 0.26 mm, 95% confidence interval (CI) 0.08 to 0.43, sd 0.38; tip, 0.11 mm, 95% CI -0.08 to 0.31, sd 0.42) compared with the conventional stem (head, 0.62 mm, 95% CI 0.34 to 0.90, sd 0.56, p = 0.02; tip, 0.43 mm, 95% CI 0.21 to 0.65, sd 0.44, p = 0.03). There was no significant difference in dynamically inducible micromotion, rate of complications or functional outcome. Conclusion This study demonstrates that the short femoral stem has a stable and predictable migration. However, longer-term survival analysis still needs to be determined. Cite this article: Bone Joint J 2018;100-B:1148–56.