We studied the survival of 8579 Charnley prostheses, in 7922 patients, according to the different types of cement that had been used. All of the patients had had a primary total hip replacement for primary coxarthrosis. The mean duration of follow-up was 3.2 years (range, zero to 6.4 years). The data were collected from the national Norwegian Arthroplasty Register. The duration of survival was defined as the time to revision due to aseptic loosening. The Kaplan-Meier estimate of survival at 5.5 years for the 1226 femoral components that had been implanted with low-viscosity cement was 94.1 per cent (95 per cent confidence interval, 92.1 to 96.2 per cent), compared with 98.1 per cent (95 per cent confidence interval, 97.5 to 98.6 per cent) for the 6589 components that had been implanted with high-viscosity cement (p < 0.0001). The remaining 764 femoral components had been implanted with Boneloc cement, which was classified as neither high nor low-viscosity, and these components were considered as a separate group in the analyses. The Boneloc cement had been used for only three years, and the two-year survival rate of these prostheses was 95.5 per cent (p < 0.0001). The cement contained an antibiotic in 2801 (42 per cent of the hips in which the femoral component had been implanted with high-viscosity cement, compared with only thirty (2 per cent) of those in which it had been implanted with low-viscosity cement.(ABSTRACT TRUNCATED AT 250 WORDS)