Cortisol Test UK: What It Can and Cannot Tell You

A cortisol test can show whether your body is making far too much cortisol or far too little — the questions behind Cushing's syndrome and adrenal insufficiency. It cannot measure how well you are coping with stress: no cortisol test has been validated for that purpose, whatever the panel is called.

The consultation is a familiar one. A woman brings a printed graph to her appointment: four saliva samples, a shaded "normal" band, one dot outside it, and underneath, in confident language, a diagnosis — written by someone who is not a doctor. She has paid for it. She is also genuinely exhausted, and nobody has taken that seriously enough. Nearly always, the graph is not the problem she came in with. Often, by the end of the appointment, there is a better answer on the table.

So this is the version worth having across a desk: which tests are validated and for what, what the timing rule really is, and where to look when cortisol is normal and the tiredness is not.

If you are here for one of the two serious ones, go straight there. Cushing's syndrome and adrenal insufficiency are the reasons cortisol testing exists, and both announce themselves as a pattern rather than as tiredness alone — the sections on each are below, with what to do. If you have severe abdominal pain with vomiting, or extreme drowsiness or confusion, stop reading and call 999. That can be an adrenal crisis, and it is an emergency.

Cortisol follows a clock, and the clock is half the test

Cortisol follows a daily rhythm — highest within an hour of waking, lowest around midnight — which is why the time of the sample changes the meaning of the number more than almost anything else. It is made by the adrenal glands, two small glands above the kidneys, on instruction from a pituitary hormone called ACTH; that chain, with its feedback loop, is the HPA axis.

Cortisol is also not a toxin. You need it to hold your blood pressure up, to keep your blood sugar steady while you sleep, and to respond when you are ill or injured — people who make too little become dangerously unwell. So the popular idea that cortisol should be flushed out of the body has the biology upside down. The aim is never less cortisol. It is the right amount at the right hour, and a test sees that only if the sample was taken at an hour where "right" has been defined.

The only three questions a cortisol test answers

Cortisol testing answers three clinical questions — too much, too little, and whether replacement dosing is right — and the fourth question it is sold for, "how are my adrenals coping", has no validated test.

Too much is the question behind Cushing's syndrome: sustained excess cortisol, most often from steroid medicines and occasionally from a pituitary or adrenal tumour. Too little is adrenal insufficiency, including Addison's disease. The third question applies only to someone already diagnosed and under an endocrinologist's care.

Then there is the fourth. Symptom scoring systems and salivary cortisol profiles have never been tested scientifically or validated as tools for evaluating the function of the HPA axis (McDermott, Journal of the Endocrine Society, 2025). That is not a complaint about assay quality; the laboratory may measure your saliva perfectly well. It means no threshold exists for "coping badly", so there is nothing to compare the result against. A number without a validated reference point only looks like an answer.

Which cortisol test answers which question

The Endocrine Society's Cushing's guideline lists four first-line tests — 24-hour urine free cortisol, late-night salivary cortisol, a 1 mg overnight dexamethasone suppression test, or a longer low-dose version of the same suppression test — and it states that random serum cortisol or plasma ACTH should not be used to screen for the condition (Nieman et al., JCEM, 2008).

The sample type matters more than most people are told. Blood measures total cortisol, most of it bound to a carrier protein called cortisol-binding globulin — so anything shifting that protein, the combined pill or pregnancy, shifts the total without shifting what is biologically active. Saliva measures the free fraction. A 24-hour collection measures free cortisol excreted across a whole day, smoothing the rhythm rather than sampling one point on it. Three different quantities, three sets of reference ranges, none of them interchangeable.

Read down the middle column and the argument makes itself. Two late-night saliva samples, collected on separate evenings between 11pm and midnight, are a validated first-line screen — validated because they are timed to catch the nadir, the hour when cortisol should have bottomed out. A four-point curve uses the same saliva, often the same laboratory, to draw a shape across the day, and that shape has never been validated against anything. So the difference is not how many samples you give. It is which hour they were taken at, and which question they were chosen to answer. The market presents the curve as the sophisticated upgrade; relative to the guideline, that is backwards.

One genuine source of confusion, worth clearing up. A cortisol day curve is a real test, and if you search the phrase you will find NHS trusts publishing patient leaflets on it. It means blood samples taken at intervals across a single day to check that a hydrocortisone dose is appropriate, with the midday and afternoon samples drawn before the next dose is due (Cambridge University Hospitals NHS Foundation Trust, Hydrocortisone Day Curve – 4 point, 2022; Manchester University NHS Foundation Trust, Cortisol Day Curve, 2024). The adrenal insufficiency guideline makes the same point at guideline level: day curve monitoring may be useful to guide dosing (Bornstein et al., JCEM, 2016). Hospital setting, blood, diagnosed patient, dosing question. A four-point saliva or dried-urine profile sold to an undiagnosed woman to explain her tiredness shares the words "day curve" and nothing else that matters: different sample, different population, different question, and no validation for the use it is sold for.

If you are holding a result and want it read properly, that is how I assess fatigue, stress and hormonal symptoms in clinic — against your timing, your medication list and your clinical pattern, not a shaded band on a printout.

What time should cortisol be tested — and what if yours was at 10am?

For a morning cortisol the sample should be early — the Société Française d'Endocrinologie anchors interpretation on a sample drawn between 8:00 and 8:30 — because a level that is reassuring at 8am may be uninterpretable by mid-morning. The same guidance sets out how that number is used when a stimulation test is not available: below roughly 140 nmol/L points to adrenal insufficiency and warrants treatment, above roughly 360 nmol/L makes it very unlikely, and anything between needs formal stimulation testing in endocrinology.

Late-night salivary cortisol works on the opposite logic. It is not looking for a peak; it is looking for a floor that never arrives. Cortisol should be at its lowest around midnight, which is why the sample is taken late. The guideline stipulates two measurements, and most clinicians collect them on two separate evenings between 11pm and midnight — the question being asked is whether the rhythm has flattened, and one evening cannot answer it. The hour is not a detail. It is the test.

Now the part nobody serves. If your blood was drawn at 10am rather than 8am, the honest answer is that the result is usually not diagnostic in either direction. It cannot reassure you and it cannot condemn you. That is an argument for repeating it properly, not for discarding the symptom — a very different conversation from being told the number is fine. One practical thing: ask for the sampling time to be written on the request form, and ask what time it was actually taken. A result with no recorded hour is very often a result nobody can interpret.

Do home and high-street cortisol tests work?

A home kit can measure cortisol accurately and still not answer your question, because what determines the answer is which test was chosen, when the sample was taken, and who interprets the result. On UK direct-to-consumer pricing checked in July 2026, a four-sample saliva panel sits at around £89, and a single blood measurement at around £42. Analytical accuracy is not the weak link. Test choice and interpretation are.

In the UK, cortisol testing for suspected Cushing's syndrome is arranged by a GP using blood, urine or saliva, with referral to an endocrinologist if the result is raised. That is the route, and it is a good one. Advertising regulators have acted against at-home test promotions implying a definitive diagnosis or discouraging medical supervision — which cuts both ways, because a stress explanation should never delay investigating something treatable.

And if you are used to the French system? Women who have moved to London from France describe a different instinct. As they tell it, you go to a laboratoire, ask for a cortisol salivaire, and collect the result yourself; the paperwork governs reimbursement rather than permission, and some say a saliva kit can simply be bought over a pharmacy counter. Set against that, London feels obstructive. The GP decides, the endocrinologist confirms, and nobody hands you a printout on request.

The instinct is not wrong. It is governed differently. But easier access to the laboratory does not fix the part that actually goes wrong: if nobody chose the right test, or pinned down the hour it was taken, you end up holding a perfectly accurate number that still cannot be interpreted. That is not a paperwork problem. It is a question problem — and it is the same reason a kit posted through your letterbox in London does not answer it either.

And if a raised result has been waved away? Ask for it in writing, with the units, the reference range and the sampling time, and ask whether a validated first-line test has been done — a 24-hour urine collection, two late-night saliva samples, or an overnight suppression test. A random daytime cortisol is not a screening test; that is guideline, not opinion. A result of 991 against a 100–600 range deserves better than a shrug.

When high cortisol is real: Cushing's syndrome

Cushing's syndrome is rare but considerably more common in women, and the pattern that warrants testing is a combination of features rather than fatigue alone. The combination that should prompt it: fat gathering on the trunk, upper back and neck while the arms and legs stay thin; a red, rounded, full face; skin that bruises from almost nothing; new wide purple stretch marks; weakness in the upper arms and thighs — trouble getting out of a low chair, or washing your hair without your arms aching. Periods often become irregular or stop. Mood changes, and libido drops.

Before any of that is investigated, prescribed steroids have to be excluded: long-term high-dose steroid medicine is by a wide margin the commonest cause of Cushing's syndrome. Inhalers, joint injections, courses of tablets, strong topical creams — they all count, and women routinely forget to mention them.

The guideline is measured about who to test: because endogenous Cushing's is uncommon, widespread testing is not encouraged, but anyone with several progressive features, or features unusual for their age, should be. That is the balance it is trying to strike, and it is the right one.

Two conditions get confused with it. PCOS, far more common, causes irregular periods, weight gain and unwanted hair growth — which is why that combination usually leads to a PCOS assessment rather than cortisol testing, unless the bruising, striae and muscle weakness are there too. Weight gain driven by insulin resistance mimics the picture too, and is better served by a weight and metabolic assessment than a saliva panel.

"Cortisol face." The full, red face of Cushing's syndrome is real and clinically recognised. The online use of the term for ordinary puffiness after poor sleep, alcohol or salt is not the same thing — and being told your face is evidence of a hormonal disorder is unkind as well as inaccurate.

When low cortisol is real: adrenal insufficiency

Adrenal insufficiency is diagnosed with a corticotropin stimulation test, not a saliva curve, and untreated it can progress to an adrenal crisis that is a medical emergency. The test uses 250 µg of synthetic ACTH, with cortisol measured at 30 or 60 minutes; a peak below 500 nmol/L (18 µg/dL) indicates adrenal insufficiency, and the exact cut-off is assay dependent (Bornstein et al., JCEM, 2016). Separately, the same guideline treats an ACTH more than twice the upper reference limit alongside a cortisol below 140 nmol/L as pointing to a primary adrenal cause. Salivary cortisol has no diagnostic role in that pathway.

It is most often diagnosed between 30 and 50, and missed for months, because the early symptoms — profound tiredness, weakness, poor appetite, low mood, dizziness on standing — overlap with almost everything else. What should prompt urgent testing is skin darker than the surrounding skin, especially in scars, creases and on the gums. Salt craving too. This is exactly the population being sold saliva panels.

Call 999 if you:

• have a fast heart rate
• feel very dizzy or light-headed (particularly when standing up after sitting or lying down)
• have severe tummy pain or pain in the side
• have muscle weakness, pain or spasms
• have a headache that does not go away
• feel sick (nausea) or vomit
• feel very drowsy, irritable or confused
• have a seizure (fit)
• lose consciousness

That list is the NHS one, reproduced as it stands, because an emergency instruction is not something to paraphrase. These are the signs of an adrenal crisis. If someone has an emergency injection and has used it, they still need emergency help — call 999 anyway. Do not wait to see whether it settles.

A normal cortisol does not close the question

A normal result does not exclude disease. In one prospective study, every one of eight patients with recurrent or persistent Cushing disease produced normal late-night salivary cortisol values on more than one occasion, and four of them did so more than half the time; six of eight newly diagnosed patients also had at least one normal value (Sandouk et al., JCEM, 2018). These were people with confirmed disease, tested with a validated test, returning normal numbers.

That finding cuts in two directions. It is why running the same unvalidated panel again because one dot looked odd achieves nothing — variation is what cortisol does, and French endocrinology guidance notes that the variability of a single cortisol measurement can reach 160 nmol/L. It is also why a symptomatic woman should not be dismissed on one normal result. Those two are not in tension. Repeating the wrong test, or the right test at the wrong hour, only adds noise; repeating the right test at the right hour is precisely what the guideline asks for, which is why both the 24-hour collection and the late-night saliva sample are specified as more than one measurement. Not a bigger panel. A better question, asked twice.

"Adrenal fatigue" is not a validated diagnosis, and here is what the evidence shows

A systematic review that screened 3,470 papers and analysed 58 studies concluded that adrenal fatigue "is still a myth", and noted that the cortisol-assessment methods those studies relied on were not endorsed by endocrinologists (Cadegiani & Kater, BMC Endocrine Disorders, 2016). There is no recognised diagnosis of adrenal fatigue, no validated test for it, and no evidence that ordinary stress wears the adrenal glands down into a state that precedes Addison's disease. It is not an early stage of anything.

The harm is not that the label is wrong. It is what the label stops. Pseudo-diagnoses do damage by convincing people they have found the explanation for their symptoms, so they stop looking for the real one — that is the argument made in the Journal of the Endocrine Society, and its consequences turn up in consulting rooms. Two years and several hundred pounds spent on "adrenal support" are two years and several hundred pounds not spent on a ferritin, a thyroid panel or a conversation about perimenopause.

Let me be careful here, because this is where these articles usually lose the reader. Your exhaustion is not imaginary. Waking at three in the morning, weight settling round your middle, feeling wrung out by a day you used to manage easily — real, and worth investigating. It is the label that has failed you, not your account of your own body. This is also what functional-medicine thinking without unvalidated tests looks like: whole-person, patient, curious about causes, and still willing to say when a test measures nothing.

If it isn't cortisol, what is it?

In practice, the treatable explanations for persistent exhaustion that get missed are iron deficiency, thyroid disease, perimenopause, obstructive sleep apnoea and mood disorder — all of which have validated tests. Each of them is routinely left sitting in a set of notes while a cortisol curve is being interpreted.

• Ferritin and iron studies. Of everything on this list, iron deficiency as a treatable cause of exhaustion is the one found most often, particularly alongside heavy periods. A ferritin in single figures explains a great deal, and a normal haemoglobin does not rule it out.
• Thyroid function. An underactive thyroid produces tiredness, cold intolerance, weight gain and low mood. Ask what your actual result was, not whether it was "fine".
• Perimenopause. Broken sleep, night sweats, mood change and shortening cycles in your late thirties and forties are far more likely than a cortisol disorder — see perimenopause in your 40s.
• Sleep. If you snore, wake unrefreshed or your partner has noticed pauses in your breathing, obstructive sleep apnoea deserves a formal assessment. And why you wake at 3am usually has a better explanation than a saliva sample.
• Mood, alcohol and depression. All three genuinely alter cortisol, which is one reason a curve is so easily over-read — and all three need treating in their own right.

What a proper assessment looks like

A four-point curve you have already paid for is worth reading rather than binning — read for what it is, which is a record of what you were told and by whom, not a diagnosis. Then the clinical question comes first: does this pattern suggest too much cortisol, too little, or neither?

If it suggests either, the route is the validated one — the right test, at the right hour, with the sampling time documented — and anything abnormal goes to endocrinology, which manages these conditions and their treatment. If it suggests neither, and that is the more common outcome, the appointment is better spent on iron studies, thyroid function, cycle history, sleep and mood. Worth knowing: the cortisol disorders that get picked up are usually recognised as a pattern before any number arrives — a cluster of physical changes that belong together, or profound fatigue alongside a sign like darkening skin. The number confirms what the history and examination have already raised.

Questions women ask

Should I get my cortisol tested? Only if the picture points to too much or too little cortisol — a combination of Cushing's features, or unexplained fatigue with darkening skin and dizziness. Testing to find out how you are coping with stress has no validated basis, so start with a consultation, not a kit.

What time should cortisol be tested? For a morning blood test, early: French endocrinology guidance interprets the result against a sample drawn between 8:00 and 8:30. For salivary cortisol, between 11pm and midnight on two separate evenings, when the level should be at its lowest. Mid-morning samples are often uninterpretable.

Do home cortisol tests work? They can measure cortisol accurately and still not answer your question, because the answer depends on which test was chosen, when the sample was taken and who reads it. A four-point saliva curve is not validated for assessing how your adrenal glands are coping.

Who can test my cortisol levels in the UK? Start with your GP, who can arrange blood, urine or saliva testing where the clinical picture warrants it, and refer to an endocrinologist if a result is raised. A gynaecologist can also assess the pattern and arrange the right tests.

Is "cortisol face" real? The full, red, rounded face of Cushing's syndrome is real and clinically recognised, but it comes with other features — easy bruising, purple stretch marks, weak upper arms. Everyday puffiness after poor sleep, salt or alcohol is a different thing, and on its own it is not a reason to test cortisol.

Is adrenal fatigue a validated diagnosis? No. A systematic review of 58 studies concluded it remains a myth, and there is no validated test for it. Your tiredness is real; the label is what fails you, because it stops the search for a cause that can be treated.

Can a cortisol test be normal and something still be wrong? Yes, in both directions. In one study, all eight patients with recurrent or persistent Cushing disease had normal late-night salivary cortisol on more than one occasion. One normal result should not end the conversation, and one abnormal result does not make a diagnosis.

Bring the result, not another kit

Three questions, three tests, and one test sold for a fourth question nobody can answer with a number. That is the whole of it. If you are holding a cortisol result nobody has explained, bring it — Dr Victoire will read it against your history and arrange the tests that are validated for what you are describing. If the exhaustion has gone on for months and no one has looked properly, what a well-woman check covers is often the better starting point.

Book a consultation in Kensington or on Harley Street.

This article is general information, not medical advice. It cannot take account of your history, your medication or your examination findings, and it is not a substitute for assessment by your GP, a gynaecologist or an endocrinologist. The thresholds quoted here are the ones clinicians work to; they are not for self-interpretation. If symptoms worry you, see a doctor — and if any of the red flags above apply, call 999.

Sources

• Cadegiani FA, Kater CE. Adrenal fatigue does not exist: a systematic review. BMC Endocrine Disorders 2016;16:48. DOI 10.1186/s12902-016-0128-4
• McDermott MT. Pseudo-endocrine disorders. Journal of the Endocrine Society 2025;9(1):bvae226. DOI 10.1210/jendso/bvae226
• Nieman LK, Biller BMK, Findling JW, et al. The diagnosis of Cushing's syndrome: an Endocrine Society clinical practice guideline. JCEM 2008;93(5):1526–1540.
• Bornstein SR, Allolio B, Arlt W, et al. Diagnosis and treatment of primary adrenal insufficiency: an Endocrine Society clinical practice guideline. JCEM 2016;101(2):364–389. DOI 10.1210/jc.2015-1710
• Sandouk Z, Johnston P, Bunch D, et al. Variability of late-night salivary cortisol in Cushing disease. JCEM 2018;103(3):983–990.
• Société Française d'Endocrinologie. Le diagnostic d'insuffisance surrénale sans Synacthène®, 2013.
• Cambridge University Hospitals NHS Foundation Trust. Hydrocortisone Day Curve – 4 point, approved 31 August 2022.
• Cambridge University Hospitals NHS Foundation Trust. Cortisol day curve, approved 22 August 2025.
• Manchester University NHS Foundation Trust. Cortisol Day Curve – Monitoring hydrocortisone replacement, v8, September 2024.
• NHS. Cushing's syndrome; Addison's disease (the "Call 999 if" list is reproduced from the Addison's disease page).