To evaluate blood transfusion practice in hip and knee arthroplasty, the development of evidence based guidelines, their implementation and prospective analysis of change. An audit was carried out in 4 stages to complete the loop. Stage 1: Retrospective analysis of blood transfusion practice in primary and revision hip and knee arthroplasty. Review of case notes, nursing record, anaesthetic sheet and pathology results from a computer database was carried out. Rates of transfusion, patients’ body weight and height, peri-and post-operative blood loss, use of anticoagulants, drains, surgical approach, type of implant and cement, grade of surgeon and anaesthetist and haemodynamic complications were recorded. Stage 2: Literature search to develop evidence based guidelines for blood transfusion. The data in stage 1 was evaluated in the light of those guidelines to determine appropriateness of blood transfusion. Stage 3: Dissemination and implementation of guidelines. Anaesthetic, Orthopaedic and audit departments were involved. Guidelines were presented, discussed, finalised and circulated. Stage 4: Prospective re-evaluation of blood transfusion practice was undertaken. Parameters as in stage 1 plus documentation of reason for blood transfusion by the prescriber were recorded. For stage 1, 97 Hip arthroplasty (86 primary and 11 revisions) and 119 Total knee arthroplasty procedures (109 primary and 10 revisions) over a period of 26 weeks were studied. Blood transfusion rate was 50.5% (49/97) in hip arthroplasty and 28.5% (34/119) in knee arthroplasty. Evidence based guidelines were developed. 55% transfusions were thought to be inappropriate in the light of guidelines. Following completion of stage 2 and 3, prospective audit of blood transfusion practice was initiated. It was compulsory for the person prescribing blood to document the indication. Data was collected on a daily basis for 15 weeks. In that period 150 joint replacements were undertaken. 77 hip arthroplasty (71 primary and 6 revisions) and 73 knee arthroplasty procedures (66 primary and 7 revisions) were undertaken. Blood transfusion rates for hip arthroplasty decreased to 18% (14/77) and for knee arthroplasty to 5.4% (4/73). Overall transfusion rates decreased from 83/216 (38.5%) to 18/150 (12%) after implementation of guidelines. This represents an overall reduction of 68%.