Cushing's syndrome tests: how doctors check for too much cortisol

Cushing's syndrome is diagnosed in steps. A doctor first checks that steroid medicine is not the cause, then arranges one or more of three first-line tests: cortisol in a 24-hour urine collection, cortisol in saliva collected late at night, and blood cortisol the morning after a dose of dexamethasone.¹ Because cortisol excess can come and go, each test is usually done at least twice.¹² If these show too much cortisol, a specialist then looks for the cause with further tests and scans.³ Hair cortisol has been studied as an alternative, with promising results, but it is not part of the diagnostic guidelines and cannot confirm or rule out the condition.

Who is tested for Cushing's syndrome?

Cushing's syndrome is rare. European population studies report two to three new cases a year of endogenous Cushing's syndrome, the kind caused by the body's own cortisol, per million people.¹ In Danish national data, the yearly rate was 1.2 to 1.7 per million for Cushing's disease (caused by a pituitary growth), 0.6 per million for benign adrenal tumours and 0.2 per million for adrenal cancer.⁴ It is much more common in women than in men, and the most common cause overall is long-term steroid medicine.³ The body's own excess can come from a pituitary growth that makes too much ACTH, the hormone that drives the adrenal glands, from a tumour elsewhere, such as in the lungs, that makes ACTH, or from an adrenal tumour that makes cortisol itself.⁵

Because the condition is rare and its common features, such as weight gain, high blood pressure and low mood, are not, the Endocrine Society recommends testing:¹

• people with several features that are getting worse, especially those that most clearly point to Cushing's: easy bruising, a flushed face, weakness of the muscles nearest the trunk, and reddish-purple stretch marks more than 1 cm wide
• people with features unusual for their age, such as thinning bones or high blood pressure
• children whose weight is rising while their growth in height slows
• people found to have an adrenal nodule on a scan done for another reason

It recommends against widespread testing of anyone else, and asks doctors to take a full medicine history first, because steroid creams, joint injections and some herbal products are easily overlooked.¹ The signs are described in more detail in high cortisol symptoms.

Finding it matters. Untreated, Cushing's syndrome raises the risk of heart attack, stroke, high blood pressure, type 2 diabetes, blood clots, osteoporosis and kidney stones,³ and in the Danish study people with non-cancerous disease died at 3.68 times the rate expected.⁴ If your symptoms may be caused by steroid medicine, do not stop it suddenly: after long-term use that can cause dangerously low cortisol.⁶

The three first-line tests

Each test looks at a different part of how cortisol is controlled. Their accuracy is described by sensitivity, the share of people with the condition a test picks up, and specificity, the share of people without it that a test correctly clears.

24-hour urine cortisol

You collect every drop of urine for 24 hours in a container from the laboratory, which measures the free cortisol passed over a whole day. Guidelines recommend at least two collections.¹ As a guide, one US reference gives normal values below 124 nmol (45 µg) per 24 hours for women and below 165 nmol (60 µg) for men, but each laboratory sets its own range.⁸ Results can be falsely low when kidney function is reduced (creatinine clearance below 60 mL/min), and sex, body size, age, urine volume and salt intake all influence them.¹² See the urine cortisol test guide for how to collect.

Late-night saliva cortisol

Cortisol normally falls to its lowest point around midnight.⁹ In healthy people with regular sleep, saliva cortisol at bedtime or between 11pm and midnight is below 145 ng/dL (4 nmol/L).¹ Published studies report 92 to 100 per cent sensitivity and 93 to 100 per cent specificity, similar to urine testing.¹ At least two samples are recommended, and the 2021 Pituitary Society consensus advises two or three.¹² The test is not suitable for night-shift workers or people with irregular bedtimes, and results can be falsely high if the swab picks up steroid cream or gel, or after liquorice, chewing tobacco, smoking or stress shortly before collection.¹² See the saliva cortisol test.

Overnight dexamethasone suppression test

Dexamethasone is a strong synthetic steroid. You take 1 mg at 11pm, and blood is taken at 8am the next morning. In a healthy person the dose switches the body's own cortisol production down, so the morning cortisol should be below 1.8 µg/dL (50 nmol/L).⁷ In Cushing's syndrome it fails to suppress, and a value above that cut-off is a positive result.¹ A longer version, 2 mg a day for 48 hours, is also used.¹ The main pitfall is oestrogen: about 50 per cent of women taking the combined pill get a falsely positive result, so oestrogen-containing medicines should be stopped for 6 weeks beforehand where possible.¹ More in our dexamethasone suppression test guide.

The 2021 Pituitary Society consensus recommends starting with any of these three tests. It states that all of them have sensitivity above 90 per cent in this setting, with lower specificity, so false positives are the bigger concern.²

Why the tests are repeated

Cortisol excess in Cushing's syndrome is often uneven. In a study of 152 people with Cushing's disease, urine cortisol varied by 52 per cent from one collection to the next in the same person, and collecting more than two samples did little to reduce that.¹⁰ Late-night saliva can swing too. When people with Cushing's disease collected it on up to 42 consecutive nights, extreme fluctuations were common, every patient with recurrent or persistent disease had at least two normal results, and the highest values in new cases ranged from 1.55 to 15.5 times the upper limit of normal.¹¹

In cyclic Cushing's syndrome, cortisol excess comes in episodes with normal periods in between. Urine and saliva tests reflect at most 24 hours, so they can be normal if collected between episodes.¹²¹ The Endocrine Society suggests urine or late-night saliva tests rather than dexamethasone tests when cyclic disease is suspected,¹ and the 2021 consensus recommends repeated, periodic late-night saliva tests to catch it.² So one normal result does not settle the question if symptoms continue.

Once the diagnosis is made and treated, the same tests are used for follow-up. The 2021 consensus recommends a late-night saliva test every year after surgery as the most sensitive way to spot a recurrence, and urine or saliva tests to monitor medical treatment.²

Hair cortisol in Cushing's research

Because hair records months of cortisol in one sample, researchers have tested whether it can detect Cushing's syndrome and capture the ups and downs that single tests miss. These are the main studies:

The authors of the 2017 study concluded that one hair sample gives diagnostic accuracy similar to the current first-line tests, not better.¹⁴ In two of their patients, whose Cushing's syndrome came from a tumour outside the pituitary, hair showed cortisol had been raised 3 and 6 months before the disease became clinically obvious.¹⁴ In the 2019 study, which used LC-MS/MS (mass spectrometry), hair cortisone, a closely related hormone measured alongside cortisol, was the most accurate measure.¹⁵ A smaller study of 36 people found hair cortisol related to urine cortisol (r = 0.5) and midnight blood cortisol (r = 0.4), and called for further validation.¹⁶

Hair's clearest potential is in cyclic disease, where a long sample can show a timeline.¹⁷ A 2026 case report described a 12-year-old boy whose conventional tests gave contradictory results. Very high cortisol along 11 cm of hair supported the diagnosis, and a pituitary tumour making ACTH was confirmed 1.5 years after he first presented, most likely with a cyclical course.¹⁸ For how segments are read, see segmental hair cortisol analysis.

Why hair is not yet a diagnostic test

Despite these results, hair cortisol is not in the main diagnostic guidelines. The Endocrine Society guideline, the 2021 Pituitary Society consensus and a 2026 consensus from the Korean and Japan Endocrine Societies do not mention it.¹²¹⁹ In 2026 a Pituitary Society expert panel of 41 participants did not endorse it as a recommended test. In the panel's first round, 61.5 per cent chose hair cortisol as an emerging biomarker that needs more research; in the second, 15.8 per cent chose it as the top research priority for monitoring disease progression and 7.7 per cent for monitoring response to treatment.²⁰ The paper's discussion suggests hair cortisol and cortisone may be the most promising way to reflect long-term cortisol excess.²⁰

There are good reasons for caution:

• It is still a research tool. A 2024 review in Endocrine Reviews said hair cortisol still needs validating outside research, with attention to sex, ethnicity, physical activity, light exposure and medicines.¹⁷
• There is no agreed cut-off. The 31.1 pg/mg threshold belongs to one study's method.¹⁴ Healthy values differ more than tenfold between methods, from 0.84 to 8.07 pg/mg in one LC-MS/MS reference study to 17.7 to 153.2 pg/mg with an early immunoassay.²¹²²
• It misses milder excess. In 67 people with adrenal nodules, hair cortisol and cortisone did not differ from healthy controls, whether or not the nodule was making cortisol on its own, and the authors judged hair unsuitable for detecting mild autonomous cortisol secretion.²³ That milder form is defined by a dexamethasone test result above 50 nmol/L without the full picture of Cushing's.²⁴
• Accuracy is similar, not better. Published sensitivity of 81 to 93 per cent and specificity of 88 to 98 per cent mean hair, like every test, would miss some cases and wrongly flag some healthy people.¹²¹⁴¹⁵

So a hair result cannot diagnose or rule out Cushing's syndrome. If a hair test has shown high cortisol, or you have symptoms whatever your result, take it to your GP, who can arrange the standard tests. For a wider look at what hair results can and cannot tell you, see is hair cortisol testing accurate?