Background Several approaches are used by blood centers when providing minor (non-ABO/D) antigen-negative RBCs to hospitals. Details vary but include providing results on the unit labeling intended for clinical use without retyping or providing results on packing documents or via computer query requiring confirmation. Recent regulatory changes allow labeling with historical minor antigen results, defined as previously performed by the donor center on two different donations with results linked to the donor and confirmed concordant. Here we investigate causes of discrepancies and identify critical process steps. Study design and methods Nine years (2009-2017) of data were reviewed for number, antigen system, and root cause of discrepancies flagged by the computer when retyping donors prior to labeling (internal discrepancies) or reported by the hospital when retested (external discrepancies). Licensed automated (CcEeK) and tube methods were used. Results Among 300,000 samples phenotyped for CcEe, K, Fya/b , Jka/b , Ss (>3 million antigens), ∼1,389,960 were repeated on 2nd donation with 397 (1/3501) discordant; 205 Fy, 118 Rh, and 74 others. Of ∼682,691 antigen-negative phenotypes provided on unit labeling, ∼37.5% (256,118) were retyped by hospitals with 29 discrepancies (1/8832), primarily Rh variants. Conclusion When repeating minor antigen types by serology, discrepancies are primarily associated with weak Fyb , among Caucasian donors, and weak/partial Rh antigens in donors of African ancestry. DNA-based testing avoids these. To label with historical results, accuracy is increased by automated testing with direct computer interface. Testing on two donations with results confirmed to be concordant is not inferior to testing on the current donation.