What an A1C or OGTT Can and Cannot Find

Written bySameer Patel — Founder & Editor

Last Updated Aug 7, 2026

Which test is actually on your lab order

Most people arrive at this question the same way. A lab order appears in a patient portal, or a sentence lands at the end of an appointment — “we’ll check your A1C,” or “come back fasting and we’ll do a tolerance test” — and it is not obvious whether one of those is the better deal.

Find yourself below.

• You have an order and no result yet. Start with what each test is looking for, then skip to what a tolerance test morning involves.
• You have a result and you are trying to interpret it. The pillar guide to reading an A1C result is the better starting point.
• Your A1C came back normal but you are still worried. The section on what each test finds is written for you, and so is the correction that follows it.
• You are pregnant or planning a pregnancy. Skip ahead — for you this stops being a preference and becomes a rule.
• You are going with someone. Read the section on what the morning involves; a tolerance test is a two-to-three-hour appointment, not a ten-minute draw.

One more piece of routing. If your glucose showed up on a routine metabolic panel rather than on a diabetes test, that is a different question with its own answer — see whether a routine metabolic panel can detect diabetes.

ℹ️ Medical Disclaimer: This article is general health education about diagnostic testing. It does not diagnose diabetes or prediabetes, does not tell you which test to have, and does not replace the judgment of the clinician who ordered your test. Decisions about which test to use, how to interpret a result, whether to repeat it, what medication to start, and what your insurance will cover are all clinical and financial decisions that belong with a board-certified primary care physician, endocrinologist, or obstetric provider who knows your history. Nothing here should be used to delay or decline testing your clinician has recommended.

What each test is actually looking for

The two tests are not fast and slow versions of the same measurement. They are looking for different things, and almost every practical difference between them falls out of that one distinction.

The A1C reads a record your blood already wrote

The A1C test measures how much glucose has attached to the hemoglobin in your red blood cells, which gives an average of your blood glucose over roughly the past three months, according to NIDDK’s guide to the tests used to diagnose diabetes — a page NIDDK last reviewed in July 2022. You do not fast for it. It uses one blood sample and can be drawn at any time of day.

Nothing is done to you during an A1C. The record already exists in your blood; the test reads it.

The tolerance test writes a new one on purpose

An oral glucose tolerance test does the opposite. You fast, a baseline sample is taken, you drink a standard sugar load, and a second sample two hours later measures what your body did with it.

NIDDK specifies the load precisely: it follows the World Health Organization’s protocol, using the equivalent of 75 grams of anhydrous glucose dissolved in water. That standardization is the point — everyone is given the same challenge, so the results mean the same thing.

🔬 How It Works: Glucose in your bloodstream binds to hemoglobin, and it stays bound for the life of the red blood cell. An A1C measures how much of that binding has accumulated, so it reports a past that has already happened and cannot be altered by what you did this morning. A tolerance test measures something that has not happened yet — it creates a controlled, measurable event and watches your body handle it. One test is a recording. The other is an experiment.

Why that difference changes what each test can catch

Because an A1C reads an accumulated record, NIDDK notes it is unaffected by short-term changes from stress or illness. Because a tolerance test measures a single morning, the same source lists it as affected by short-term lifestyle changes including stress, illness and medications.

That cuts both ways, and the trade it creates is the subject of the next section. If your two numbers have already come back and they disagree with each other, that is a different question — see why an A1C and a fasting glucose can disagree.

NIDDK also states a general interpretive principle worth carrying through everything below: all laboratory test results represent a range rather than an exact number. That is true of the A1C, and it is true of glucose. What a healthy day-to-day reading looks like is covered separately in what a normal blood sugar reading looks like; this page stays on the numbers that define a diagnosis.

The test that finds the most is not the steadiest one

There is a federal comparison of these tests that almost no patient ever sees, and it answers the question this page is about more directly than any consumer article does.

A federal comparison ranks the three tests, and the common one comes last

NIDDK publishes a side-by-side comparison of the diabetes blood tests, last reviewed August 2020. On sensitivity — how many true cases a test picks up — it states the relationship three times, consistently, and the ranking is the same each time.

The tolerance test is more sensitive than the A1C or the fasting glucose; the fasting glucose is more sensitive than the A1C; and the A1C is less sensitive than both.

The same page lists this among the A1C’s drawbacks in plain terms: lower sensitivity, identifying fewer cases of diabetes than the glucose tests do.

That is a federal source saying the most convenient test finds the fewest cases. It is also only half of what that page says.

The same page ranks them the other way on consistency

Read the other columns and the ranking inverts. NIDDK lists the tolerance test as having low reproducibility and high within-patient variability. It lists the A1C as having low within-patient variability, the most stable sample after collection, established international standardization of laboratory tests, and monitored accuracy.

So the more sensitive test is the less repeatable one. Run a tolerance test twice on the same person in the same month and you are more likely to get two meaningfully different answers than you would with two A1C tests.

Neither test wins. They fail in opposite directions, and which failure matters more depends on what the test is being asked to do.

Source: NIDDK, “Diabetes & Prediabetes Tests,” last reviewed August 2020. Both columns come from the same federal comparison.

Critically, being less sensitive does not make the A1C less valid. Under the American Diabetes Association’s Standards of Care in Diabetes—2026, Section 2, diabetes may be diagnosed on A1C criteria or on plasma glucose criteria — which include the 2-hour value during a 75-gram tolerance test — so these are alternatives rather than a ranking of validity.

A high A1C is a diagnosis on its own terms, and a reader who leaves this page doubting a real elevated result has read it wrong — what an A1C result can and cannot settle explains why a diagnosis gets confirmed rather than dismissed.

What it means if your 2-hour result sits exactly on the line

📊 Clinical Data Point: On a 2-hour tolerance test, a result of 140 to 199 mg/dL indicates impaired glucose tolerance and 200 mg/dL or above indicates diabetes, repeated to confirm — Source: NIDDK, “Diabetes & Prediabetes Tests,” August 2020, and “Diabetes Tests & Diagnosis,” July 2022. ⚠️ The two federal pages render the boundary differently: CDC’s page on diabetes testing, which displays a date of May 15, 2024, lists normal as 140 mg/dL or below and prediabetes as 140 to 199; NIDDK’s patient table stops normal at 139 mg/dL or below. A result of exactly 140 sits where the two public renderings differ.

This is a difference in how the same underlying criterion is written out, not a claim that either agency is wrong. But it has a practical consequence: if your 2-hour number came back at exactly 140, you cannot resolve your own case by reading a table, because the two tables you are most likely to find disagree at that single value.

✅ Patient Action: If your 2-hour result was 140, ask the clinician who ordered it — your primary care physician or the ordering provider — this specific question: “My 2-hour result was right on the line. Does that change what you would do next, or do you want to repeat it?”

Why the A1C is usually the one you are offered

The honest answer to “which one and why” is that most non-pregnant adults in the US do not choose. The A1C is what arrives, and there are real clinical reasons for that rather than cost-cutting ones.

Convenience is a clinical property, not a shortcut

NIDDK lists the A1C as requiring no fasting, sampled at any time of day, convenient for both patient and clinician, and the most stable sample after collection. It lists the tolerance test as not convenient for either party, requiring fasting and a scheduled morning appointment or return visit, involving extensive patient preparation, with a sample that is not stable after collection.

A test that is easy to complete gets completed. A test that requires a fasting morning and a two-hour wait gets rescheduled, skipped, or abandoned partway through — and a test nobody takes finds nothing.

NIDDK’s patient page adds that the tolerance test is more expensive than the fasting glucose test and the glucose challenge test, and not as easy to give. Note the scope of that comparison carefully: it is against those two tests, not against the A1C. If cost is your actual question, what an A1C test costs handles it properly rather than in one clause.

What the newest federal page says the tolerance test is for

Here two authoritative sources describe the same test differently, and understanding why is the answer to the reader’s question.

The ADA Standards and NIDDK’s comparison describe what the tolerance test can do: it is a co-equal diagnostic criterion with the highest sensitivity of the three. MedlinePlus’s page on what happens during a glucose test, last updated December 30, 2024 and the newest of these sources, describes what it is mostly used for: diagnosing gestational diabetes, and — if you are not pregnant — sometimes used to confirm a diagnosis of prediabetes or type 2 diabetes.

Those two statements are not in conflict. One is a statement of validity, the other a statement of practice. The test is fully valid and it is not routine, and both are true at once.

🩺 Physician Note: The direction of confirmation is not fixed either. MedlinePlus describes the tolerance test as sometimes used to confirm a diagnosis outside pregnancy — while NIDDK’s A1C page states that in some cases clinicians use the A1C test to help confirm the results of another blood glucose test. Each test is described as the other’s confirmation, by two different federal pages. Neither one is subordinate; which comes first depends on which one raised the question.

The reasons a doctor adds the second test

A tolerance test tends to be added rather than substituted, and the reasons are specific. Pregnancy is one, and it is covered in the next section. The others are something about your blood making the A1C unreliable, or a result close to a threshold that needs settling.

NIDDK also notes that diagnosis requires two abnormal results, from the same sample or from two different samples. What that pair of results actually settles is its own subject — see what a second test actually settles. And if you are already sitting in the prediabetes range, what doctors check after a prediabetes-range result picks up where this page leaves off.

✅ Patient Action: Before your next appointment, ask the ordering provider one question: “Is there anything about me — my blood, my medications, a pregnancy — that would make an A1C less reliable here?” That is a better question than asking for a specific test by name.

When the tolerance test is the right test for you

This section applies to a minority of readers. If none of it describes you, that is the expected outcome and not a reason to doubt the test you were given.

In pregnancy this is not a preference

Here the answer stops being a judgment call. NIDDK states plainly on its page on the A1C test, last reviewed April 2018, that the A1C test should not be used to diagnose gestational diabetes. The glucose challenge test or the oral glucose tolerance test are used to check for it, usually between 24 and 28 weeks of pregnancy.

NIDDK’s patient page adds the surrounding schedule: everyone without a previous diagnosis should be tested for gestational diabetes between 24 and 28 weeks, and testing for type 2 diabetes should happen at the first prenatal appointment or within the first 15 weeks of pregnancy. If a glucose challenge test comes back too high, a return visit for a fasting tolerance test may follow. In pregnancy, blood is drawn every hour for two to three hours rather than once at two hours.

⚠️ Clinical Warning: If you are pregnant, do not treat a normal or reassuring A1C as a gestational diabetes screen. NIDDK is explicit that the A1C is not used to diagnose gestational diabetes at all — an early-pregnancy A1C is looking for diabetes that predated the pregnancy, and it reflects months that may include time before you conceived. Substituting it for the 24-to-28-week screen means the condition that screen exists to find would not be looked for.

NIDDK adds one point that is easy to miss after delivery: if you had gestational diabetes, you should be tested for diabetes no later than 12 weeks after your baby is born.

When something about your blood makes the A1C the wrong instrument

The A1C measures glucose attached to hemoglobin, so anything that changes your hemoglobin or the lifespan of your red blood cells can change the result without your glucose having changed at all.

NIDDK’s comparison lists an altered relationship between A1C and glycemia in a number of conditions, and lists possible interference from certain genetic hemoglobin variants and elevated fetal hemoglobin. It also states that the A1C is not recommended for rapidly progressing diabetes such as type 1 diabetes in children, and not recommended for screening cystic fibrosis-related diabetes.

Two things are worth saying about that list rather than reproducing it. First, most of it does not apply to most people. Second, the full explanation of why and how these interfere belongs elsewhere — what can make an A1C result unreliable covers the mechanism, and sickle cell trait and A1C testing covers the single most common example in detail.

What to ask, and what not to conclude

The usable signal here is relational, not numerical. NIDDK notes that clinicians may suspect interference when your A1C and your blood glucose results do not match each other.

That is something you can check on your own records without a threshold: do your numbers agree, or do they tell different stories? What you should not do is scan a list of conditions looking for a reason to set aside a result — that reverses the logic, and it is the most common way this information gets misused.

✅ Patient Action: If your A1C and your glucose readings have not matched, bring that to your primary care physician or, in pregnancy, your obstetric provider, and ask: “My A1C and my glucose numbers don’t line up. Is there a reason an A1C would be unreliable for me, and if so, which test would you use instead?”

What a glucose tolerance test morning involves

Most of the anxiety on this question is procedural rather than clinical. Here is what the appointment actually is.

The fast, and what counts as breaking it

You must fast for at least eight hours before a tolerance test, according to both NIDDK and MedlinePlus. Fasting means nothing to eat or drink except sips of water.

The A1C requires no fasting at all and can be sampled at any time of day — that contrast is one of the clearest practical differences between the two, and whether you need to fast for an A1C covers the A1C side in full.

What you eat in the three days before matters too

This is the instruction most consumer pages omit. NIDDK’s comparison specifies that patients should take in at least 150 grams per day of carbohydrates for three days before a tolerance test.

That is a real preparation requirement, not a suggestion, and it means a deliberate low-carbohydrate week before your appointment is not neutral preparation. NIDDK specifies the intake; it does not state which way an inadequate intake moves the result, so neither will this page.

🩺 Physician Note: Follow the preparation instructions your laboratory gives you, and tell the lab if you have been eating very differently from usual in the days beforehand — a low-carbohydrate diet, a fasting protocol, or an illness that changed what you could keep down. The three-day carbohydrate requirement exists because the test measures a response, and a response depends on the state your body is in when it is tested.

How long you are actually there

Plan for the morning, not for a quick stop. A fasting sample is taken, you drink the glucose liquid, and a further sample is taken two hours later. MedlinePlus describes samples being taken about every hour for the next two or three hours; NIDDK describes hourly draws across two to three hours in pregnancy.

You stay at the facility throughout. That is the part people are most often unprepared for, and it is why a caregiver coming along should expect a two-to-three-hour wait rather than a ten-minute draw.

Why you might want someone to drive you home

MedlinePlus states that after an oral glucose tolerance test you may feel light-headed, and that your provider may suggest you plan to have someone take you home.

That single sentence appears almost nowhere in the consumer coverage of this test, and it is the most practically useful thing on this page for anyone booking an appointment. Arrange the ride before the day, not during it.

CDC adds one more point that stops a common mistake: results can differ depending on the size of the glucose drink and how often your blood sugar is tested. Two people describing “my tolerance test result” may not have taken the same test, which is a reason not to compare your number against a friend’s.

What neither test can tell you

Two limits are worth knowing before you decide either result has settled something.

Neither one tells you which type of diabetes you have

NIDDK is direct about this: the tests that confirm diabetes cannot identify what type it is — type 1, type 2, or a rare form called monogenic diabetes. Genetic testing and autoantibody testing are what distinguish them.

That matters because type is what drives treatment. A reader who has been arguing with themselves about A1C versus tolerance test is often really asking a question neither test answers.

There is a second thing to resist. If the sensitivity comparison above left you intending to demand a tolerance test, reconsider the framing — asking whether an A1C is reliable for you is a good question that a clinician can act on, while requesting a specific test by name is not the same thing and is easier to decline.

Why there is no at-home version of a tolerance test

Disclosure: this site earns advertising revenue, and we are declining an obvious affiliate category here. We do not recommend at-home or direct-to-consumer testing kits as a way of getting either of these tests done, and the reasons are on the record rather than editorial preference.

NIDDK states that blood testing equipment you can buy over the counter, such as a blood glucose meter, cannot diagnose diabetes. It also states that samples analyzed in a doctor’s office or clinic — point-of-care tests — should not be used for diagnosis; ADA’s 2026 Standards restrict point-of-care A1C for screening and diagnosis to devices approved for diagnosis by the FDA, used in laboratories certified to perform moderate-complexity or higher testing. CDC tells anyone whose blood sugar was tested at a health fair or pharmacy to follow up with their doctor, who may want to re-test.

And there is a structural reason specific to this test. There is no at-home oral glucose tolerance test — it requires timed draws under supervision across two to three hours, with a standardized load. A mail-in kit cannot produce it, and selling one beside an article explaining a supervised challenge test would misrepresent the thing being explained.

If your question is really about getting a test without going through a clinician, ordering a blood panel online without a doctor covers what that route can and cannot do.

One last piece of context on what a normal result buys you. NIDDK’s guidance is that adults and children with normal results should be retested every three years, and that those diagnosed with prediabetes should be tested for type 2 diabetes every year. A normal result is a result for now, not a permanent clearance.

Common questions about the A1C and the tolerance test

What to do with whichever test you are given

The tolerance test finds more cases; the A1C repeats more consistently; and under the 2026 Standards either can diagnose diabetes. That is a trade between two valid instruments, not a ranking, and it is why the test on your order is usually the one that fits your circumstances rather than the one that was cheaper.

So the useful step is not to switch tests. Turn up prepared for the one you were given — fasting and hydrated if it is a tolerance test, any time of day if it is an A1C. And if anything in the section on reliability applies to you, bring that specific fact to your ordering provider and ask whether it changes which test they would use.

How this was made

About this content

How this article was put together: researched from recognised health sources, drafted with the help of AI tools, and edited by hand, with sources linked throughout.

1 contributor

Written byView profile

Sameer Patel — Founder & Editor

Researched and written from recognised health sources

Sameer Patel is the founder and editor of My Medicine Advisor. He is not a doctor or medical professional — before starting this site he worked in banking,…

Content on MyMedicineAdvisor is researched from recognised sources, drafted with AI assistance, and edited by hand, with sources linked so you can check them. It is general health education, not medical advice, and there is no clinical review panel.