Abstract Background Accurate values for hemoglobin A1c (HbA1c) are crucial for diagnosing and monitoring diabetes mellitus. Two common methods to measure HbA1c are capillary electrophoresis (CE) and immunoassay (IA). Many methods used to measure HbA1c are prone to interferences from hemoglobin variants and abnormalities affecting the red blood cell’s (RBC) turnover. Consequently, the average estimated glucose levels using the HbA1c value can be unreliable in diabetics with hemoglobinopathies and other RBC lifetime disorders. Separation methods such as CE can capture hemoglobinopathy information, while IA methods may not. We performed this study in a population with relatively high levels of hemoglobin disorders to evaluate the performance of HbA1c testing on the Alinity platform (IA) against the Sebia Capillarys 3 Tera instrument (CE). Methods Samples were identified retrospectively based on a search of the Electronic Medical Record for all patients with an ordered and resulted HbA1c and complete blood count in the Montefiore Medical Center the day prior to sampling. Routine HbA1c testing was performed by an Abbott Immunoassay method on the Abbott Alinity Instrument and CBC results were obtained from a Sysmex XN2100 instrument. Records were compared by patient name, medical record number, and sample collection time to identify all patients with CBC and HbA1c drawn within one hour of each other. Those samples were retrieved from storage (a 4° automated refrigerator) within 48 hours of draw and tested via the Sebia Capillarys 3 Tera instrument using HBA1c kits. Results In this large representative population study (N=2075), there was overall excellent agreement between IA and CE. The correlation coefficient (R2) of the whole dataset was 0.989, with an average bias of 2.7% between the two methods. Additional analysis was performed on those with possible RBC turnover disorder (patients with RBC counts below 4x106/uL or above 6x106/uL, MCVs above 96 fL or below 80 fL, RDW above 17%, or abnormal A2 value: total of 638 samples), those with a hemoglobin variant as detected by the Capillarys 3 instrument (162 samples), or with a measurable hemoglobin F value detected by the Capillarys 3 instrument (37 samples). Samples were sorted into these groups independently and thus there was significant overlap between groups, notably with 23 of 37 samples with elevated HbF also having a hemoglobin variant. In both the RBC disorder and hemoglobinopathies group, similar low bias and high correlations were seen. The elevated F samples showed a clear negative bias (5.6%) for all patients on the IA method as compared with CE. The IA method states that samples with HbF above 5% should not be tested. However, as this is not a separation method, there is no way to test samples for HbF before analysis, and bias was seen for samples below 5%. Conclusions The IA and CE methods showed similar performance across patients regardless of RBC lifetime or hemoglobin disorder status, though presence of HbF is a potential interference and should be considered. Further questions remain for these instruments in terms of correlation with average glucose levels as calculated through CGM instruments.