Because levels of glycosylated hemoglobin (GHb) are increased in diabetes and reflect the previous metabolic control, clinicians and clinical investigators are finding increasing applications for measurements of GHb in diabetic patients. We report the characterization of a colorimetric assay procedure for GHb and compare its performance with that of a commonly used assay by ion-exchange chromatography. Although results of GHb determination by both methods correlate highly (r = 0.946, P less than 0.001), the two procedures estimate different glycosylated fractions. The colorimetric procedure is nonstoichiometric, requiring careful standardization of assay conditions, including the concentration of total hemoglobin in the assayed aliquot, to achieve precision and permit comparison of results. We characterized the effect of storage of hemolysates or packed erythrocytes on the subsequent determination of GHb by both methods. Determinations of GHb by the colorimetric method, but not by column chromatography, are reproducible on hemolysates or packed erythrocytes on the subsequent determination of GHb by both methods. Determinations of GHb by the colorimetric method, but not by column chromatography, are reproducible on hemolysates or packed erythrocytes stored frozen for at least 5 mo. A unique advantage of the colorimetric procedure is the capability to estimate GHb levels when variant hemoglobins, including fetal and sickle hemoglobins, are present.