Background: A Group O Policy, which issues group O red blood cells (O-RBC) to recipients whose ABO types have not been duplicated on at least 2 separate occasions, has been in place at this 550-bed hospital since 1991. Its inception was a response to the frequency of wrong blood in tube (WBIT) events identified by discrepant serial ABO typings, which had been disproportionately concentrated among critically ill patients, a population at high risk of transfusion and misdraw. This policy was thought to better address such sources of mistransfusion error than delivery-based patient identification technologies.Methods: For the period from 1/1/2004 - 12/31/2005, the policy's impact was reviewed according to utilization of O-RBC, frequency of critical O-RBC shortage, intervals to type confirmation, the identification of protracted users of policy O-RBC, the frequency of WBIT events as revealed by ABO typing discrepancies, the frequency of ABO-incompatible transfusion errors, and the frequency with which ABO-incompatible transfusion errors were averted as a result of the policy.Results: In 2004, 479 (4.5%) of a total of 10,575 units of O-RBC were transfused due to the policy. In 2005, 689 (6.1%) of a total of 11,293 units of O-RBC were transfused due to the policy. Increase in the use of O-RBC from 2004 to 2005 was by 718 units (6.7%), while the increase in the use of group O-RBC for the policy increased disproportionately by 210 units (43.8%). Over the 2 year period, a monthly median excess of 43 units (range 24–85) of O-RBC were issued in accordance with the policy. Daily excess use of O-RBC for the policy was a median of 1 unit (range 0–19). There were 291 days when no policy-related O-RBC were used, and only 3 days when more than 10 units were used. The interval between a 1st and 2nd (confirmatory) typing, when achieved, occurred within 24 hours for 15.4% of patients, but was a median of 23 days, and ranged from minutes to 21.5 years. Confirmatory re-typing activity within less than 72h of the 1st occurred in a minority of 28%, possibly reflecting nonessential repetition and/or deliberate action to release patients from policy-related O-RBC restrictions. Patients who were protracted users of Group O policy blood (no confirmatory type and/or repeatedly indeterminate typing) were neonates (310 units [26.5%]) and recipients of ABO-incompatible hematopoietic stem cell transplants (76 units [6.5%]). The blood bank identified misdrawn blood through typing discrepancies in 6 patients over 2 years. Transfusion occurred uneventfully in 3 of these 6, as the misdraw was made apparent by comparison with prior typings on file in 2, and patient knowledge of correct type in 1. The misdrawn specimen was the 1st one in 2 of 6. Of the 3 who were not transfused, the misdraw suggested a type that would have led in all to the transfusion of clinically incompatible blood, had the knowledge of prior type, and the policy, both been absent. ABO-incompatible red cell transfusion errors did not occur, although this was a greater function of luck than the policy in the interval.Conclusions: A policy instituted to prevent ABO-incompatible transfusion errors, by providing O-RBC to those with unconfirmed blood types, appeared to be both feasible, and potentially mistransfusion-preventative, at a high-volume tertiary-care teaching hospital.