Hemoglobin A1c (HbA1c) is an important index of average glycemia in patients with diabetes mellitus that is widely used in clinical trials and large-scale epidemiological studies. Previous studies have shown that adverse sample storage conditions can cause erroneous HbA1c results. We examined the effect of storage at different temperatures with five current HbA1c methods: Tosoh G7 and G8 (Tosoh Bioscience, Inc., South San Francisco, CA) and Bio-Rad Variant™ II (Bio-Rad Laboratories, Hercules, CA) (all ion-exchange high-performance liquid chromatography); Siemens DCA 2000+ (Siemens Healthcare Diagnostics, Deerfield, IL) (immunoassay); and Trinity Biotech (Kansas City, MO) ultra(2) (boronate-affinity high-performance liquid chromatography).Five whole blood specimens with different HbA1c levels were analyzed by each assay method on Day 0 and then divided into aliquots that were stored at six different temperatures (-70°C, -20°C, 4°C, room temperature, 30°C, and 37°C) for analyses on subsequent days out to Day 84. Acceptance limits were defined as within ±3 SD of all -70°C results or ±0.2% HbA1c, whichever was wider, for each sample. Stability was considered acceptable for a given temperature only if results for all five specimens were acceptable on that day.The DCA 2000+ demonstrated the best stability at -20°C and room temperature, whereas the ultra(2) showed the best stability with specimens stored at 4°C. No methods demonstrated stability at 30°C or 37°C for more than 3 days.Exposure of specimens to high temperatures should be avoided regardless of assay methodology. For the ion-exchange methods tested 4°C storage is preferable to -20°C (stability 14-21 days vs. 4-10 days). For studies where long-term stability is required, samples should be stored at -70°C or colder.