OBJECTIVE To evaluate the safety and efficacy of a preoperative autologous blood donation (PABD) program with liberal patient exclusion criteria. METHODS Prospective patient accrual from September 1990 to May 1992. METHODS A PABD program in a public tertiary care hospital linked with a specialist orthopaedic practice in a private hospital. METHODS One hundred and twenty patients consecutively scheduled for orthopaedic surgery were referred for PABD. There were 77 women (median age, 63 years) and 43 men (median age, 62 years). One-third of patients were 70 or more years old. Coexistent medical disease, mainly cardiovascular, was present in 46% of patients and 3% were excluded because of coexisting morbidity. Total hip or total knee replacement was performed in a private hospital with intraoperative or postoperative blood salvage in 37% of the patients. METHODS Adverse effects of PABD, autologous blood collected and used and homologous blood transfused. RESULTS One hundred and sixteen patients donated 267 units of autologous blood. In 70 patients undergoing total hip replacement, 78% donated three and 20% donated two units, with 95% of autologous blood being used. In 38 patients undergoing total knee replacement, 42% donated two units and 55% donated one unit, with 87% of autologous blood being transfused. Seventy-eight per cent of all patients only received autologous blood. In the remaining patients, homologous blood use was confined to two units or less in 80%. Hypotensive episodes associated with phlebotomy occurred in 3% of patients, and 13% of patients had preoperative haemoglobin levels ranging from 85-100 g/L without adverse clinical effects. CONCLUSIONS A hospital-based PABD program with less strict patient exclusion criteria does not prejudice the clinical status of the donors. Its linkage to private hospitals can significantly reduce the use of homologous blood in selected elective surgery.