The correct wristband identification of patients is essential to prevent acute, hemolytic transfusion reactions from incompatible transfusion. We compared wristband identification errors for 712 hospitals. Phlebotomists checked patient wristbands on 2,463,727 occasions, finding 67,289 errors; in 33,308 instances, patient wristbands were missing entirely. The median total error rate was 2.2%; 10% of participants had error rates of 10.9% or greater. Absent wristbands represented 49.5% of errors; multiple wristbands with different information, 8.3%; wristbands with incomplete data, 7.5%; erroneous data, 8.6%; illegible data, 5.7%; and patients wearing wristbands with another patient's identifying information, 0.5%. The monitoring for errors by phlebotomy staff was the most important policy associated with lower error rates. Initial placement of wristbands by nursing staff was the only policy associated with increased error rates. We conclude that wristband identification error rates depend on differences in hospital policy and procedure and should be responsive to quality improvement efforts.