Background and purpose Elderly patients with displaced femoral neck fractures are commonly treated with a hemiarthroplasty (HA), but little is known about the long-term failure of this treatment. We compared reoperation rates for patients aged at least 75 years with displaced femoral neck fractures treated with either internal fixation (IF), cemented HA, or uncemented HA (with or without hydroxyapatite coating), after 12-19 years of follow-up. Methods 4 hospitals with clearly defined guidelines for the treatment of 75+ year-old patients with a displaced femoral neck fracture were included. Cohort 1 (1991-1993) with 180 patients had undergone IF; cohort 2 (1991-1995) with 203 patients had received an uncemented bipolar Ultima HA stem (Austin-Moore); cohort 3 (1991-1995) with 209 patients had received a cemented Charnley-Hastings HA; and cohort 4 (1991-1998) with 158 patients had received an uncemented hydroxyapatite-coated Furlong HA. Data were retrieved from patient files, from the region-based patient administrative system, and from the National Registry of Patients at the end of 2010. We performed survival analysis with adjustment for comorbidity, age, and sex. Results Cemented HA had a reoperation rate (RR) of 5% and was used as reference in the Cox regression analysis, which showed significantly higher hazard ratios (HRs) for IF (HR = 3.8, 95% CI: 1.9-7.5; RR = 18%), uncemented HA (HR = 2.2, CI: 1.1-4.5; RR = 11%) and uncemented hydroxyapatite-coated HA (HR = 3.6, CI: 1.8-7.4; RR = 16%). Interpretation Cemented HA has a superior long-term hip survival rate compared to IF and uncemented HA (with and without hydroxyapatite coating) in patients aged 75 years or more with displaced femoral neck fractures.