The extensive benefits of the total hip (THA) and knee (TKA) replacements are well documented, but surprisingly little is known about their economics. We assessed costs, cost-effectiveness (C/E), and patient-related C/E variances in THA and TKA from data on 276 THA and 176 TKA patients. Patients with primary arthrosis, primary operation, and total joint replacement were recruited from seven hospitals between March 1991 and June 1992. Their use of health and other welfare services together with health-related quality of life (HRQoL) were measured before the surgery and at 6, 12, and 24 months postoperatively. HRQoL was assessed by the 15D, a 15-dimensional HRQoL instrument, and the Nottingham Health Profile. Costs were assessed from questionnaire responses, the Finnish Hospital Discharge Register, and Finnish Arthroplasty Register. Total hospital costs per patient were 45,000 FIM (US $10,500) for THA and 49,600 FIM (US $11,500) for TKA. Prosthesis costs comprised 21% of these costs in THA and 24% in TKA. On average, hip patients gained more in terms of HRQoL, and the operations were more cost-effective. The C/E ratio for younger (< or = 60 years) knee patients did not differ from those in all age groups of hip patients, whereas TKAs in those over 60 years had a worse C/E ratio compared with all other patient subgroups. It was concluded that allocation efficiency can be improved by considering not only the intervention but also patient characteristics such as age. Indeed, the C/E ratio varied more across age groups of knee patients than between average THA and TKA patients.