Primary uncemented femoral stems reported to the Norwegian arthroplasty register between 1987 and 2005 were included in this prospective observational study. There were 11 516 hips (9679 patients) and 14 different designs of stem. Kaplan-Meier survival probabilities and Cox regression were used to analyse the data. With aseptic loosening as the end-point, all currently used designs performed excellently with survival of 96% to 100% at ten years. With the end-point as stem revision for any cause, the long-term results of the different designs varied from poor to excellent, with survival at 15 years ranging between 29% and 97%. Follow-up for longer than seven years was needed to identify some of the poorly-performing designs. There were differences between the stems; the Corail, used in 5456 hips, was the most frequently used stem with a survival of 97% at 15 years. Male gender was associated with an increased risk of revision of x 1.3 (95% confidence interval 1.05 to 1.52), but age and diagnosis had no influence on the results. Overall, modern uncemented femoral stems performed well. Moderate differences in survival between well-performing stems should be interpreted with caution since the differences may be caused by factors other than the stem itself.