Shortages of donor blood and fears of transmitted infections have prompted the use of a range of blood-sparing techniques in the peri-operative period. We conducted a systematic review of three techniques that involve the re-infusion of a patient's blood--pre-operative autologous blood deposit (PAD), acute normovolaemic haemodilution (ANH), and cell salvage (CS). We examined the effects of these interventions on the need for peri-operative allogeneic red blood cell transfusion and on clinical outcomes. Controlled clinical studies were identified by computer searches of comprehensive electronic databases and bibliographic searches of published articles. The literature search retrieved a total of 68 randomized trials (RCTs) and 81 controlled observational studies that included data from over 34 000 individuals. In summary, the RCTs found that autologous transfusion techniques consistently reduced the frequency of allogeneic transfusions, with intervention effect sizes ranging from a relative 63% reduction (95% CI 46-74%) with PAD, to 42% (27-53%) with CS and to 31% (16-44%) with ANH. Non-randomized studies reported larger effect sizes than RCTs. PAD increased overall transfusion rates by 30% (95% CI 12-48%) and reduced pre-operative haemoglobin levels by an average of 1.23 g dL(-1). Intervention effects were substantially reduced when these techniques were performed under transfusion protocols. Interpretation of the studies was hampered by serious methodological weaknesses, particularly inadequate randomization techniques, unblinded measurements and the subjective nature of the outcome variables. The studies reported few clinical outcome and adverse event data. Previous claims regarding reduced rates of mortality and infection with autologous transfusions were not confirmed.