The reference range for healthy adults is 4.8–5.9%.
The decision limits for nonpregnant adults, according to the American Diabetes Association, are as follows:
For patients with diabetes mellitus, the goal of therapy is less than 7.0%.
The diagnostic criterion for diabetes is greater than or equal to 6.5% NGSP units.
Hemoglobin A1c (glycated hemoglobin) reflects the average blood glucose concentration over the course of the RBC lifespan, roughly 120 days in normal individuals.
It provides different, and complementary, information to a single glucose concentration. Some patients may have near normal fasting glucose values but very high postprandial levels, and others may have elevated fasting levels with only moderately elevated postprandial levels. Hemoglobin A1c provides information comparable to what might be provided by having frequent glucose values throughout the day over the course of 3 months. [1, 2, 3]
Thus, elevated values give a sense of the degree of overall glucose control in patients with diabetes mellitus. Intensive glucose control in diabetic patients, reflected in lower hemoglobin A1c values, has been shown to "delay the onset and slow the progression of diabetic retinopathy, nephropathy, and neuropathy."  The goal of therapy is to attain a value of less than 7.0% (while minimizing hypoglycemic episodes).
Hemoglobin A1c should be monitored regularly in diabetic patients.
As of January, 2010, the American Diabetes Association began promoting the use of hemoglobin A1c as the preferred diagnostic test for diabetes mellitus. Among its advantages over fasting glucose values (or 2-hour glucose values during an oral glucose tolerance test) is that samples can be drawn at any time and need no special handling (whereas ongoing glycolysis can falsely lower glucose values). [5, 6]
Most laboratories report a calculated eAG (estimated average glucose) along with every measured hemoglobin A1c, which is designed to facilitate communication with patients as well as to help clinicians appreciate the degree of hyperglycemia the A1c represents.
Note that falsely low values can occur in patients whose RBC lifespan is shorter than normal (eg, hemolytic anemia, some hemoglobinopathies) as well as in patients who have received transfusions from nondiabetic patients.
Hemoglobin A1c is a specific fraction of hemoglobin A found in healthy individuals as well as individuals with diabetes mellitus. It is formed when the N-terminal valine of the beta chain of hemoglobin A is modified by the addition of a sugar moiety. This process is nonenyzmatic, so the term glycation is preferred to the term glycosylation, although the latter is commonly used. Once formed, hemoglobin A1c is stable. As a result, the hemoglobin A1c level reflects the average blood glucose level over the course of the red blood cell’s lifespan, roughly 120 days.
A landmark paper showed convincingly that intensive control of diabetes (ie, maintaining near-normal concentrations of glucose throughout the day, reflected in lower A1c levels) "delays the onset and slows the progression of diabetic retinopathy, nephropathy, and neuropathy."  Fingerstick glucose measurements by patients remain the mainstay of diabetes management for adjusting daily insulin doses, but this study established the importance of ongoing, periodic A1c measurements to monitor compliance and efficacy of therapy.
For example, in patients whose records of fingerstick glucose are incomplete, reflecting good control in the week prior to a physician’s visit, a hemoglobin A1c value of 6.8% is reassuring, whereas a value of 9.6% indicates that glucose levels were probably much higher in the preceding weeks.
A limitation of hemoglobin A1c is that it does not provide any indication of the changes in glucose concentrations throughout the day, for which frequent glucose measurements are needed. In the image below, both patients would have the same A1c level, but the patient in blue has far fewer highs and lows, and represents a much better degree of control.
The extent of hemoglobin glycation is related not only to the glucose concentration in blood but also to the average RBC lifespan. In patients with shortened average lifespans, the hemoglobin A1c level can be misleading (falsely low). [7, 8] Similarly, in patients who have recently been transfused, the hemoglobin A1c level will reflect, to some extent, the donors’ glucose levels.
With the relatively recent introduction of a reference method for the measurement of hemoglobin A1c,  it became apparent that the methods used historically, though very highly correlated to the true hemoglobin A1c concentrations, included chemical species other than A1c; in fact, the values were roughly 2 units higher than the true value (ie, a value of 7% was closer to 5% in conventional units).  Different countries have approached this problem in different ways. Many countries have adopted totally new units (mmol/mol), but the United States has decided to continue to report conventional units (%).  The relationship between the 2 sets of units is as follows:
IFCC-A1c (mmol/mol) = [NGSP-A1c (%) – 2.15] x 10.929
However, in an effort to smooth the transition to the new units, as well as to help both physicians and patients appreciate the clinical context of various hemoglobin A1c levels, many laboratories are now reporting eAG (estimated average glucose) along with every hemoglobin A1c level.  Although the correlation is far from perfect and the use of this parameter has been somewhat controversial,  it has received support from many major organizations, including the American Diabetes Association, the American Association for Clinical Chemistry, and the College of American Pathologists. The equation to calculate eAG (in mg/dL) from hemoglobin A1c (in %) is as follows:
eAG (mg/dL) = 28.7 x NGSP-A1c (%) – 46.7
On balance, clearly, reporting an eAG of 226 mg/dL in addition to an A1c of 9.5% can enhance patient care.
In addition to its use in monitoring therapy in patients with diabetes, hemoglobin A1c has recently been touted as having a role in screening for diabetes. Indeed, as of January, 2010, the American Diabetes Association has promoted the assay in its practice guidelines as the preferred test (using a criterion of ≥6.5% NGSP units). 
A short note is in order with respect to hemoglobin variants. As long as one normal beta chain exists (eg, sickle cell trait, or hemoglobin C trait), one can measure hemoglobin A1c. Even in some cases with no normal beta chains (eg, sickle cell disease, or SC disease), the abnormal beta chains are glycated and can be measured by some methods. In other cases (eg, hereditary persistence of fetal hemoglobin), no beta chains exist to glycate, so glycated hemoglobin is not present. In all of these cases, though, the RBC lifespan issue may take precedence over the ability (or the lack thereof) to measure glycated hemoglobin and must be kept in mind. As noted earlier, regular fingerstick glucose monitoring throughout the course of the day can provide reliable data in these situations.