A 71-year-old woman, weighing 70 kg, presented for surgery to revise first the right and then the left femoral and acetabular cemented prostheses with uncemented hydroxy-apatite ceramic coated Furlong® components (Joint Replacement Instrumentation, London). Her medical history was of essential hypertension, stable angina pectoris and an uncomplicated myocardial infarction 5 years previously. There was no evidence of cardiac failure or cerebrovascular disease. Medications included metoprolol, glyceryl trinitrate spray and oral ferrous sulphate for 2 months before surgery. On examination her blood pressure was 130/70 mmHg with no signs of cardiac failure and no carotid bruits. Electrocardiogram revealed a left bundle–branch block, and the chest radiograph showed unfolding of the aorta with a normal cardiac shadow and lung fields. Urea and electrolytes were within normal limits and haemoglobin (Hb) was 141 g/litre (operation 1) and 140 g/litre (operation 2). The patient expressed a desire not to receive any blood or blood-related products at any time, and completed the requisite Jehovah's Witness documentation. It was agreed that she would undergo preoperative isovolaemic haemodilution and intraoperative salvage of red blood cells, maintaining continuity with her circulation at all times. On both occasions the patient was premedicated with oral temazepam 20 mg and a 5 mg glyceryl trinitrate skin patch. After inserting a radial arterial line, a total of 700 ml of blood was venesected into bags containing citrate for anticoagulation and kept in continuity with the patient's circulation. Blood pressure was maintained during venesection using two 3 mg ephedrine boluses before volume replacement with 1000 ml of Gelofusine® (Braun, Melsungen) was commenced. Anaesthesia was induced with thiopentone 200 mg and maintained with isoflurane and nitrous oxide in 40% oxygen. Muscle relaxation was achieved using vecuronium and the patient intubated after receiving alfentanil 0.5 mg. Her lungs were mechanically ventilated to an end tidal PaCO2 of 4.8–5.0 kPa. Intraoperative and postoperative analgesia was achieved on both occasions with a lumbar epidural catheter inserted at L1/2 interspace, using a continuous infusion of bupivacaine 1 mg/ml and fentanyl 2 μg/ml. Standard intraoperative monitoring included invasive blood pressure, central venous pressure and oesophageal temperature. A urethral catheter was inserted and body temperature maintained around 36°C using intravenous fluid warmers and a Warm Touch® (Gaymar, Hamburg). Intraoperative red blood cell salvage was achieved with a Haemonetics Cell Saver®5 (Haemonetics Corporation, Massachusetts), set up with a continuous circuit from the surgical sucker, through the filtration and reservoir units to an intravenous line in the patient using normal saline as a primer. Intraoperative systolic blood pressure was maintained between 90–100 mmHg. Surgery involved meticulous removal of all cement using cement splitting and extracting instruments under image intensifier control. Finally granulation tissue between cement and bone was curetted away to expose bleeding cancellous bone. There were no unexpected surgical complications. Intraoperative blood loss totalled 1250 ml (operation 1), and 1100 ml (operation 2). A total of 2000 ml of colloid was infused after isovolaemic haemodilution and before the completion of surgery, at which time an arterial blood gas sample revealed no significant metabolic acidosis on either occasion. After wound closure a total of 500 ml of Cell Saver®5 infusate with an estimated haematocrit of 0.55 was transfused. Finally the two 350 ml bags of autologous blood were infused slowly over the next few hours. Following extubation the patient was transferred to intensive care for overnight monitoring and continuous postoperative epidural analgesia. Investigations included full blood count, urea and electrolytes, coagulation profile, and arterial blood gases. Serial 12-lead electrocardiograms and cardiac enzymes revealed no evidence of myocardial ischaemia or infarction on either occasion. Demonstrable postoperative blood losses totalled approximately 500 ml for each operation. The lowest recorded Hb was 80 g/litre (operation 1), and 89 g/litre (operation 2), measured intraoperatively before infusing the Cell Saver®5 infusate. Figure 1 shows the Hb levels from the preoperative day through to discharge for both procedures. Aside from a 37% prolongation of prothrombin time and 27% prolongation of activated partial thromboplastin time on the first postoperative night, coagulation rapidly returned to normal values. The patient recovered uneventfully and despite persistent anaemia (Figure 1), was discharged home after 1 week on both occasions. Two-year follow-up confirms successful revision arthroplasties.