Background and purpose - Previous studies of patients who have undergone total hip arthroplasty (THA) due to femoral head necrosis (FHN) have shown an increased risk of revision compared to cases with primary osteoarthritis (POA), but recent studies have suggested that this procedure is not associated with poor outcome. We compared the risk of revision after operation with THA due to FHN or POA in the Nordic Arthroplasty Register Association (NARA) database including Denmark, Finland, Norway, and Sweden. Patients and methods - 427,806 THAs performed between 1995 and 2011 were included. The relative risk of revision for any reason, for aseptic loosening, dislocation, deep infection, and periprosthetic fracture was studied before and after adjustment for covariates using Cox regression models. Results - 416,217 hips with POA (mean age 69 (SD 10), 59% females) and 11,589 with FHN (mean age 65 (SD 16), 58% females) were registered. The mean follow-up was 6.3 (SD 4.3) years. After 2 years of observation, 1.7% in the POA group and 3.0% in the FHN group had been revised. The corresponding proportions after 16 years of observation were 4.2% and 6.1%, respectively. The 16-year survival in the 2 groups was 86% (95% CI: 86-86) and 77% (CI: 74-80). After adjusting for covariates, the relative risk (RR) of revision for any reason was higher in patients with FHN for both periods studied (up to 2 years: RR = 1.44, 95% CI: 1.34-1.54; p <0.001; and 2-16 years: RR = 1.25, 1.14-1.38; p <0.001). Interpretation - Patients with FHN had an overall increased risk of revision. This increased risk persisted over the entire period of observation and covered more or less all of the 4 most common reasons for revision.