Thyroid-stimulating hormone (TSH)

TSH stands for thyroid-stimulating hormone. Its older scientific name is thyrotropin, which is why papers and guidelines often use that word.

The thyroid is a butterfly-shaped gland at the front of the neck. It makes T4, the main hormone it releases, and a smaller amount of T3, the active form. Tissues convert T4 to T3 as they need it.

The American Association of Clinical Endocrinologists and the American Thyroid Association (ATA) call serum TSH the single best screening test for primary thyroid dysfunction in most outpatient settings [1]. Primary means the problem sits in the thyroid itself, which covers the large majority of cases.

A TSH test does not measure thyroid hormone directly. It measures how hard the pituitary is pushing the thyroid. That is why the result moves in the opposite direction to thyroid hormone:

• High TSH usually means the thyroid is underactive (hypothyroidism), so the pituitary pushes harder. See what a high TSH means.
• Low TSH usually means the thyroid is overactive (hyperthyroidism), or that someone takes more thyroid hormone than they need. See what a low TSH means.

In the US NHANES III survey of 17,353 people aged 12 and over, 4.6% had hypothyroidism (0.3% overt, 4.3% subclinical) and 1.3% had hyperthyroidism (0.5% overt, 0.7% subclinical), most of them undiagnosed [9].

How the test developed

Utiger's 1965 radioimmunoassay first measured TSH in human plasma [10]. Early assays could not tell a low TSH from a normal one.

Spencer's 1990 chemiluminometric assay reached a functional sensitivity of about 0.018 mU/L, roughly 100 times better than first-generation assays, and the authors called it the first third-generation TSH assay [11]. Once low values were measurable, studies such as Framingham could link low TSH with atrial fibrillation [5].

NHANES III then described TSH across the US population [9], which set off the 2005 debate over the upper limit [12][13] and the 2007 age-specific analysis [8]. The most recent milestones are trials and guidelines: TRUST [7], the BMJ panel [14] and the 2026 ATA pregnancy guideline [15].

This loop is called the hypothalamic-pituitary-thyroid axis. It works like a thermostat:

1. The hypothalamus senses thyroid hormone and releases thyrotropin-releasing hormone (TRH).
2. TRH tells the pituitary to release TSH.
3. TSH binds to the thyroid and drives it to make and release T4 and some T3.
4. Rising T4 and T3 signal back to the pituitary and hypothalamus to release less TRH and TSH.

Why TSH is so sensitive

In 1990, Spencer and colleagues used a new, much more sensitive TSH assay to plot TSH against free T4 in healthy people. They found an inverse log-linear relationship: free T4 changes in a straight line while TSH changes on a logarithmic scale [11]. In plain terms, a modest dip in free T4 can multiply TSH several times over.

That amplification is why TSH often turns abnormal before free T4 leaves its own reference range. A raised TSH with a normal free T4 is called subclinical hypothyroidism; a low TSH with normal free T4 and T3 is subclinical hyperthyroidism.

Each person has a set point

When Danish researchers tested 16 healthy men every month for a year, each man's TSH moved within a band about half as wide as the population range [16]. A single TSH result described a person's own set point only to within about plus or minus 50%, and a change of about 0.75 mU/L (range 0.2 to 1.6) was needed before a repeat result could be called a real change [16].

For the full picture of how TSH, free T4 and T3 relate, read TSH vs free T4.

Overt disease matters

Untreated hypothyroidism can cause fatigue, weight gain, memory and concentration problems, menstrual changes, infertility and heart failure; in rare severe cases it progresses to myxedema coma, which carries a mortality of up to 30% [2]. Hashimoto thyroiditis, an autoimmune disease, causes up to 85% of primary hypothyroidism where iodine intake is adequate [2].

Hyperthyroidism affects about 2.5% of adults worldwide and is linked to arrhythmias, heart failure, osteoporosis, adverse pregnancy outcomes and higher mortality; Graves disease is the most common cause [17].

Low TSH and the heart and bones

• In the Framingham study of 2,007 adults aged 60 and over, a TSH of 0.1 mU/L or lower carried a 28% ten-year incidence of atrial fibrillation versus 11% with normal TSH, a relative risk of 3.1 [5].
• Pooled data on 52,674 adults linked endogenous subclinical hyperthyroidism with higher total mortality (hazard ratio 1.24) and atrial fibrillation (1.68), with the highest risks below 0.10 mIU/L [18].
• Among 70,298 adults in 13 cohorts, TSH below 0.10 mIU/L carried a hazard ratio of 1.61 for hip fracture and 3.57 for spine fracture [6].

High TSH and the heart

Pooled data on 55,287 adults found no excess coronary risk at TSH 4.5 to 6.9 mIU/L, but an 89% higher risk of coronary events at 10 to 19.9 mIU/L [19]. Hypothyroidism also raises cholesterol: treating overt hypothyroidism lowered LDL cholesterol by about 41 mg/dL and ApoB by about 34 mg/dL in a meta-analysis [20]. That is one reason a thyroid check often comes up when LDL cholesterol or ApoB is unexpectedly high.

Treating mild changes: the TRUST trial

The TRUST trial randomized 737 adults aged 65 and over with persistent TSH of 4.60 to 19.99 mIU/L to levothyroxine or placebo. Levothyroxine brought mean TSH to 3.63 mIU/L versus 5.48 with placebo, yet symptom and tiredness scores did not differ at one year [7]. A meta-analysis of 21 trials with 2,192 adults reached the same conclusion for quality of life and thyroid symptoms [21], and a BMJ guideline panel then recommended against routine treatment for most adults with subclinical hypothyroidism [14].

Labs set their own reference intervals from their own assay and population, so always read your result against the range printed beside it. Large studies commonly use about 0.45 to 4.49 mIU/L as normal [19][18], and the ETA describes mild subclinical hypothyroidism as TSH 4.0 to 10.0 mU/L [4]. Note that 1 mIU/L is the same as 1 µIU/mL, and mU/L means the same thing.

The table below lines up the main thresholds. Sources: ETA [4][22], NICE [3], AACE and ATA [1], the BMJ panel [14], pooled cohorts [19][18][6], NHANES III [8], the ATA pregnancy guideline as summarized by the Consortium on Thyroid and Pregnancy [23][24].

Four rules that change how a number reads

• One abnormal result is a starting point. The ETA asks for a repeat TSH and free T4 after 2 to 3 months before calling a raised TSH subclinical hypothyroidism [4], and NICE bases treatment decisions on two tests 3 months apart [3]. In older adults with one raised TSH, 60.8% returned to normal within about a year without treatment [25].
• Free T4 decides overt versus subclinical. A high TSH with low free T4 is overt hypothyroidism; with normal free T4 it is subclinical [2][26].
• Age shifts the range. The upper limit rises with age (see the chart in what a normal TSH is) [8].
• Pregnancy has its own ranges. Trimester-specific ranges from the local population are the standard [23][15].

None of this replaces a clinician's reading. Diagnosis, and any decision about levothyroxine or its dose, belongs with the person who knows your history.

This grid compiles patterns described across guidelines and reviews. The cascade itself comes from NICE: measure TSH first, add free T4 when TSH is high, and add free T4 and free T3 when TSH is low [3]. Overt hyperthyroidism means suppressed TSH with high T3, high free T4, or both [17]. Overt primary hypothyroidism means high TSH with low free T4 [2].

The mismatched patterns deserve care:

• Critical illness often produces low T3, low T4 and a normal or slightly low TSH [27]. During recovery TSH can rise above normal for a while before T4 catches up [28].
• Biotin supplements at 10 mg a day distorted 9 of 23 biotin-based immunoassays, pushing competitive assays such as free T4 falsely high and sandwich assays such as TSH falsely low [29]. That can look exactly like hyperthyroidism.
• Medicines such as glucocorticoids, dopamine agonists and somatostatin analogues can suppress TSH at the pituitary [30].

The repeat interval in the table follows the ETA for raised TSH [4] and NICE for low TSH [3].

Sources for the table: time of day [31][13], meals [32], illness [27][3], recovery [28], biotin [29], lithium [33], amiodarone and checkpoint inhibitors [2], pituitary-acting drugs [30], iodine [34][2], age [8].

Time of day and sleep

TSH rises in a nightly surge. Surks and colleagues note that early morning values run higher than later ones and that sleep deprivation, strenuous exercise and night or evening shifts accentuate this [13]. If you track TSH over time, draw it at a similar time each visit. A steady sleep schedule helps make repeat results comparable; the sleep consistency protocol covers the basics.

Illness

NICE advises against testing thyroid function during an acute illness unless the illness itself may be thyroid-related [3].

Biotin

NICE asks clinicians to check biotin intake because high-dose supplements can give falsely high or low results [3]. Biotin is common in hair, skin and nail products.

Age

The chart compares TSH upper limits in young and old adults from three countries [8][35][36]. In iodine-rich Korea the range was higher at every age and did not climb with age [34].

The case for a narrower range

Wartofsky and Dickey argued in 2005 that older reference populations included people with unrecognized Hashimoto disease, which pushed the upper limit up. They cited laboratory guidance that more than 95% of normal people have TSH below 2.5 mU/L, and noted a mean TSH of 1.18 mU/L in Black Americans, a group with a low rate of Hashimoto disease [12].

The case for keeping the current range

Surks and colleagues replied that 85% of the NHANES III reference group was already below 2.5 mIU/L, and that lowering the limit would relabel the 9.7% between 2.5 and 4.5, about 20.6 million Americans, as diseased. Many of them have no thyroid disease, and no benefit from treatment has been shown [13]. They suggested rechecking TSH every 1 to 2 years instead [13].

Where the evidence points now

• Age matters more than a single target. TSH rises with age in people without thyroid disease, and age-specific ranges reduce diagnoses of subclinical hypothyroidism in older adults [8][37].
• Treatment trials do not support chasing a lower number in mild disease [7][21].
• Personal set points are real. Each person's TSH sits within a narrow personal band [16], which keeps the idea of an individual target alive in research, though no test yet tells you your set point in practice.
• For people already on levothyroxine, the ETA suggested aiming for the lower half of the reference range (0.4 to 2.5 mU/L) for most adults [4]. That target is set and adjusted by a clinician, never by changing a dose on your own.

See what a normal TSH level is for ranges by age, sex and pregnancy.

• Younger adults with mild subclinical hypothyroidism. TRUST enrolled people aged 65 and over [7]. The BMJ panel noted its recommendation may not apply to young adults, such as those 30 and under, or to people with severe symptoms [14].
• Hard outcomes from treating subclinical hypothyroidism. Thyroid hormones may have little or no effect on cardiovascular events or mortality, but the evidence is low quality and harms were measured in only one trial [14].
• Treating subclinical hyperthyroidism. Fracture and heart risks are clear at TSH below 0.10 mIU/L, but whether treatment prevents fractures is still untested [6], and the ETA guideline was written without randomized trials [22].
• Pregnancy shortcuts. Fixed limits and adjusted nonpregnant ranges missed many women who needed treatment compared with trimester-specific ranges [38].
• Optimal ranges by ethnicity. TSH runs lower in Black Americans than in white Americans [9][39], and what that means clinically is still open [9].
• Persistent symptoms on treatment. Some people on levothyroxine with a normal TSH still report symptoms. Thyroid researchers such as Antonio Bianco argue that TSH alone may not capture hormone effects in tissues; this remains an open research question.