Saliva cortisol test: the different types, how to collect and how to read the results

A saliva cortisol test measures cortisol in a small sample of saliva, collected by spitting into a tube or chewing on a swab. Saliva carries the free, active fraction of cortisol, and its level follows the free cortisol in your blood within minutes.¹ There is no needle, and you can collect at home at the times that matter, including late at night or the moment you wake. That makes saliva the standard way to look for a raised late-night cortisol when Cushing's syndrome is suspected, and a popular way to see the shape of your cortisol across a day. Its weakness is the one shared by every single-moment test: a sample reflects the hour, and the day, on which you took it.

What saliva cortisol measures

In the blood, about 90 per cent of cortisol is bound to a carrier protein called corticosteroid-binding globulin (CBG), and only about 4 to 6 per cent is free.² Only the free cortisol passes into saliva. Salivary cortisol is directly proportional to free cortisol in the blood, and the two come into balance in well under 5 minutes, so a saliva sample is a close and quick reflection of the active hormone.¹

Because it ignores the bound fraction, saliva is not thrown off by changes in CBG.² That makes a practical difference for anyone taking oral oestrogen, in many contraceptive pills and some HRT. In one study, women on oral oestrogen had much higher total blood cortisol than women who were not, but their salivary cortisol was similar, at 13.8 to 15.7 nmol/L across all three groups studied.³

The three kinds of saliva test

"Saliva cortisol test" covers three quite different tests, and it is worth knowing which one you are being offered.

A daily profile

Several samples are taken across one day, typically on waking, around midday, in the late afternoon and at bedtime, to show the shape of your daily curve. In healthy people who sleep at night, cortisol peaks within three hours of waking and falls to its lowest point around midnight.² Kits of this kind are sold direct to the public; the table lists examples. A daily profile is not one of the tests the clinical guidelines use to diagnose Cushing's syndrome or adrenal insufficiency.⁴⁵ Saliva tests are also sometimes offered for "adrenal fatigue", but the Endocrine Society says no test can detect it, because there is no scientific proof that it is a real medical condition.⁶ See adrenal fatigue.

The cortisol awakening response

The cortisol awakening response is the sharp rise in cortisol over the first 30 to 45 minutes after waking.⁷ In three studies of 152 people, from children to older adults, free cortisol rose by 50 to 75 per cent within 30 minutes of waking.⁸ A review put the typical rise at 50 to 160 per cent, about 9 nmol/L on average, and found no agreed norms for absolute levels.⁹ Because the result depends on sampling from the exact moment of waking, an international expert panel has published guidelines on getting the timing right.⁷ It is mainly a research measure. One day's response is largely set by the circumstances of that day, and two to six days of sampling are needed for a reliable picture of the person.¹⁰

The late-night test for Cushing's syndrome

This is the saliva test doctors use. In healthy people cortisol reaches its lowest point around midnight,² so a raised late-night level is a warning sign of Cushing's syndrome, in which the body makes too much cortisol. The Endocrine Society guideline asks for samples on two separate evenings between 11pm and midnight.⁴ In healthy people with regular sleep, a result at that time is below 145 ng/dL (4 nmol/L).⁴ Laboratories set their own ranges: one UK hospital uses below 3.2 nmol/L for a late-night sample and 3.2 to 22.8 nmol/L for an early-morning one.¹¹

Published studies give the late-night test a sensitivity of 92 to 100 per cent and a specificity of 93 to 100 per cent for Cushing's syndrome, similar to the 24-hour urine test.⁴ Sensitivity is the share of people with the condition the test picks up; specificity is the share without it who correctly test normal. A 2021 consensus advises at least two or three late-night samples, and avoiding the test in night-shift workers.¹² It also recommends a yearly late-night saliva test after surgery for Cushing's disease, as the most sensitive way to spot a recurrence.¹² See testing for Cushing's syndrome.

How to collect a saliva sample

Follow the instructions with your kit exactly, because each laboratory has its own. The general rules are these.

• Nothing in your mouth beforehand. Do not eat, drink, brush or floss your teeth for at least 30 minutes before a sample; some kits ask for longer.¹³ A protein-rich meal reliably raises salivary cortisol, so food matters for more than contamination.¹⁴
• Get the time right. Note the exact time of each sample. For the awakening response, the first sample is taken at the moment of waking.⁷
• Rest first. Stress and exercise raise cortisol, so collect when you are calm and have not just been active.¹³
• Medicines and creams. Tell the person ordering the test about every medicine you use, including skin creams, and do not stop any medicine without talking to them first.¹³

What can spoil a result

Saliva tests are sensitive to things that would not affect a blood test. The Endocrine Society guideline lists these causes of falsely high results.⁴

• Steroid-containing lotions, creams or mouth gels getting onto the swab or tube.
• Liquorice and chewing tobacco.
• Smoking on the day of collection.
• Stress just before collecting.

Timing matters too. The late-night test assumes a normal sleep pattern, so it is not suitable for shift workers or people with variable bedtimes.⁴¹² If you work nights, tell the doctor before you collect, and see cortisol and shift work.

Why one day of saliva is not enough

A saliva sample is a snapshot, and most of what it captures is the day itself. In a classic study of students sampled at the same time of day six weeks apart, about 75 per cent of the variation in salivary cortisol reflected the situation on the day, about 21 per cent the person's stable level and about 4 per cent measurement error.¹⁵ Researchers who modelled how many days are needed found that a reliable average takes samples on 3 days, the area under the daily curve 4 to 8 days and the daily slope 10 days.¹⁶ They recommend fewer samples per day spread over more days.¹⁶ Across three studies of healthy adults, about half the variation in the awakening response, the daily slope and total daily output was day-to-day fluctuation.¹⁷

The same is true in disease. In people with Cushing's disease who collected late-night saliva on up to 42 consecutive nights, extreme fluctuations were common, and every patient with recurrent or persistent disease had two or more normal results.¹⁸ That is why guidelines ask for more than one sample before drawing conclusions,⁴¹² and why a single normal daily profile from a home kit cannot rule anything out.

Reading your results

Compare each value with the range on your report for that time of day; ranges are set locally and differ between laboratories.¹¹ For a daily profile, look at the shape as well as the numbers: a typical curve is highest in the morning and lowest around midnight.² There are no agreed norms for the awakening response.⁹

A raised late-night result is a reason to see a doctor, who will usually repeat it and add other first-line tests, such as a 24-hour urine collection or a dexamethasone suppression test.⁴¹² Our guides to the urine cortisol test and the dexamethasone suppression test explain them. A result outside the range from a home kit is not a diagnosis, and a result inside it does not rule out a condition if you have symptoms.

Saliva or hair: which answers your question?

If a doctor suspects Cushing's syndrome, late-night saliva is one of the guideline tests, and hair is not a substitute: hair cortisol is not part of the diagnostic guidelines.⁴¹² If your question is about the long term, whether your cortisol has been running high or low for months, hair is the more consistent measure. In a study of 17 adults who collected saliva three times a day for 30 days, cortisol in the centimetre of hair nearest the scalp correlated with their 30-day salivary average at r = 0.61.¹⁹ The same people's hair cortisol correlated at r = 0.84 from one month to the next, against 0.38 to 0.61 week to week for saliva.¹⁹

A hair test, in effect, does the averaging that would otherwise take weeks of saliva tubes, though it cannot show the shape of a single day. See hair vs saliva cortisol tests for the full comparison.