What Is a Normal Body Temperature — and When to Worry?

The famous 98.6 °F (37.0 °C) benchmark, drilled into generations of parents and patients, is not quite right. Modern research puts the average oral temperature closer to 98.2 °F (36.8 °C), and the upper end of normal at roughly 99.9 °F rather than the 100.4 °F long used as the fever cutoff.¹ That shift matters more than it sounds, because where you draw the line between “fine” and “fever” changes what you do next. And “normal” itself turns out to be a moving target shaped by time of day, age, sex, how you take the reading, and even which century you happen to live in.

Why 98.6 °F Was Never Quite Right

The 98.6 °F standard traces back to Carl Wunderlich, a German physician who published a massive dataset in 1868 based on armpit readings from thousands of patients. His work was groundbreaking for the era, but it had limitations: the thermometers were slow and less precise, the measurement site (axillary) runs cooler than oral, and the population was largely hospitalized. When researchers revisited the question with modern oral thermometers in the early 1990s, they found the actual mean was about 36.8 °C (98.2 °F), not 37.0 °C.¹ Half a degree Fahrenheit may not seem like much, but when you are deciding whether a child’s 99.5 °F reading counts as a fever, it changes the conversation.

Even more striking, human body temperatures appear to have been gradually declining over the past two centuries. A large analysis of American health records spanning birth decades from the early 1800s through the late 1990s found a steady drop of about 0.03 °C per decade in both men and women.² The leading theory is that reduced chronic inflammation, thanks to better sanitation, antibiotics, and lower rates of lingering infections like tuberculosis, has nudged baseline temperatures down over generations. Whatever the cause, it means a person born today likely runs a bit cooler than a person born a century ago.

What “Normal” Actually Looks Like

A systematic review pooling data from dozens of studies found that the overall mean oral temperature, plus or minus two standard deviations, falls between about 35.7 °C and 37.4 °C (96.3–99.3 °F).³ That is a surprisingly wide band, and it gets wider or narrower depending on where you stick the thermometer:

• Rectal: roughly 36.3–37.8 °C (97.3–100.0 °F)
• Tympanic (ear): roughly 35.8–37.5 °C (96.4–99.5 °F)
• Oral: roughly 35.7–37.4 °C (96.3–99.3 °F)
• Axillary (armpit): roughly 35.0–36.9 °C (95.0–98.4 °F)

Rectal readings consistently run highest because they reflect core body temperature most closely, while armpit readings run lowest because the sensor sits on the skin surface. Those differences are baked into the biology, not a flaw in the thermometer. If you switch measurement sites, expect the numbers to shift accordingly.

Your Temperature Is Not the Same All Day

Body temperature follows a reliable daily rhythm. It bottoms out in the early morning, roughly two hours before you wake up, and peaks in the evening, usually between about 6 PM and 10 PM.⁴ A large-scale analysis of wearable-device data confirmed this cycle, showing that the daily low typically lands between 6 AM and 8 AM and the daily high between 6 PM and 8 PM.⁵ The swing from trough to peak averages about 0.5 °C (0.9 °F), which means a perfectly healthy person can read 97.5 °F at dawn and 98.9 °F after dinner, with both numbers being entirely normal.

This daily rhythm explains a common worry: your child feels warm at bedtime, you take a temperature, and it reads 99.3 °F. At that hour, in the context of evening physiology, that number may well be normal. The same reading at 6 AM would be more meaningful.

For people who menstruate, there is an additional layer. After ovulation, rising progesterone shifts baseline temperature upward by about 0.3–0.7 °C (roughly 0.5–1.3 °F) during the second half of the cycle.⁶ This is the same shift that fertility-awareness methods rely on when tracking basal body temperature. The correlation between progesterone and temperature is moderate but consistent: progesterone and temperature tend to rise together, though temperature plateaus once progesterone reaches a certain level.⁷

Where and How You Measure Matters More Than You Think

Rectal temperatures have long been considered the clinical gold standard, particularly for young children, because they best approximate core temperature. In children older than one month, armpit readings tend to run about 1 °C (1.8 °F) lower than rectal readings on average.⁸ Oral readings fall in between, averaging about 0.25 °C below rectal in one pediatric study.⁹

The pandemic-era explosion of non-contact infrared thermometers (NCITs), the ones pointed at your forehead at airport checkpoints and clinic doors, introduced a new layer of uncertainty. A study evaluating multiple NCIT models found that, depending on the device, anywhere from about half to nearly 90 percent of individual readings fell outside the manufacturer’s own stated accuracy range.¹⁰ In a clinical comparison, contactless forehead thermometers had a sensitivity of only about 13 percent for detecting fevers above 38 °C, meaning they missed the vast majority of fevers.¹¹ Temple and forehead skin-surface readings showed essentially no agreement with core temperature in another evaluation.¹²

If you are screening a crowd for obvious high fevers, a contactless scan is better than nothing. But if you are trying to determine whether a specific person actually has a fever, an oral or tympanic thermometer is far more reliable. For infants, rectal remains the recommended route.

Age Changes the Equation

Older adults tend to run cooler at baseline and mount weaker fevers when they are sick. In a study of nursing home residents, the average oral baseline temperature was only about 97.4 °F, and during confirmed infections, nearly half the episodes produced a maximum temperature below 101 °F.¹³ About a quarter of those “blunted” fevers actually represented a meaningful temperature change from baseline, but because the baseline was so low to begin with, the reading never crossed the traditional threshold. This pattern held during the COVID-19 pandemic as well: advancing age was associated with both lower baseline temperatures and lower peak temperatures during SARS-CoV-2 infection in nursing home populations.¹⁴

The practical implication is that waiting for an elderly person to hit 100.4 °F before suspecting infection can mean missing the infection entirely. For older adults, a temperature change of 2 °F or more above their personal baseline may be a more useful alarm than any fixed cutoff.

At the other end of life, fever in very young infants demands immediate attention. Clinical guidelines recommend that any infant under 28 days old with a fever be hospitalized for evaluation and antibiotic treatment, because their immune systems are immature and serious bacterial infections can progress rapidly.¹⁵ For babies 28 to 90 days old, low-risk criteria can allow outpatient management, but close follow-up is still essential.

What Fever Actually Does

Fever is not the illness itself. It is a coordinated defense response. When the immune system detects an infection, it triggers a signaling cascade that ultimately produces a molecule called prostaglandin E2 (PGE2). That molecule acts on temperature-regulating neurons in the brain to raise the body’s thermostat setpoint.¹⁶ This is what makes fever different from overheating: during fever, the body actively defends the higher temperature the same way it normally defends 98.6 °F, shivering to generate heat and constricting blood vessels in the skin to prevent heat loss.¹⁷ That is why you feel cold and shivery at the onset of a fever, even though your temperature is actually climbing.

The distinction between fever and hyperthermia matters clinically. In hyperthermia, such as heatstroke, the body’s thermostat has not changed; it is simply overwhelmed by external or internal heat that it cannot shed fast enough. Standard fever-reducing drugs work on fever (they block prostaglandin production) but do nothing for true hyperthermia, because the problem is not the setpoint. Heatstroke requires external cooling.

Fever has been conserved across vertebrates for over 600 million years, which is strong indirect evidence that it provides a real survival advantage.¹⁸ Even cold-blooded animals generate a version of it: fish with bacterial infections will swim to warmer water to raise their body temperature, and research in Nile tilapia showed that this “behavioral fever” improved the ability of immune cells to kill pathogens and reduced immune-cell death during infection.¹⁹

When Fever Becomes a Worry

For most healthy adults with a common infection, fever in the range of 100–103 °F is uncomfortable but not dangerous. The worry thresholds depend on who you are:

• Healthy adults: A temperature above 103 °F (39.4 °C) that persists warrants medical attention, and anything above 105 °F (40.6 °C) is a medical emergency.
• Infants under 3 months: Any rectal temperature of 100.4 °F (38 °C) or higher calls for prompt evaluation.
• Older adults: As noted above, an elderly person may be seriously ill at 100 °F if their baseline runs in the low 97s.
• Immunocompromised people: A lower threshold for concern applies, because a blunted immune response may not generate a full fever even when infection is severe.

One of the most anxiety-provoking complications of childhood fever is febrile seizures, which occur in up to about 4 percent of children between 6 months and 5 years old.²⁰ They are terrifying to witness but, reassuringly, a large long-term study found essentially no differences in intelligence, academic progress, or behavior at the 10-year mark between children who had febrile convulsions and those who did not.²¹ Simple febrile seizures, the kind that last under 15 minutes and do not recur within 24 hours, are the most common type and carry an excellent long-term prognosis.

When Running Too Cold Is the Real Danger

While most attention focuses on high temperatures, an abnormally low temperature during illness can be more ominous. In sepsis, patients whose temperatures dropped below normal within the first 24 hours had dramatically worse outcomes: 28-day mortality was about 50 percent in the hypothermic group versus about 25 percent in those who were not hypothermic.²² A meta-analysis of clinical trials in septic patients confirmed the pattern: hypothermia was associated with higher mortality compared to fever.²³ The likely explanation is not that the low temperature itself is harmful but that it signals a body too overwhelmed by infection to mount a proper fever response. It is the body shifting into energy-conservation mode when the fight is going badly.

This is worth knowing because it runs counter to the common assumption that only high temperatures are dangerous. If someone with a known infection feels cold and clammy and their temperature reads below 96 °F (35.6 °C), that warrants urgent medical attention just as much as a 104 °F reading.

Fever Phobia and Overtreating Normal Temperatures

A substantial body of research documents widespread parental anxiety about fever that is out of proportion to the actual risks. In one well-known survey, over half of caregivers said they were “very worried” about fever’s potential harm, and about a fifth believed fever could cause brain damage.²⁴ A quarter of the caregivers in that study were giving fever-reducing medication at temperatures below 100 °F, and 85 percent said they would wake a sleeping child to give medicine. More recent surveys from other countries found a similar pattern: roughly a third of parents started giving antipyretics before the child’s temperature even reached 100.4 °F.²⁵

The problem with aggressive fever treatment at low temperatures is twofold. First, you may be medicating a number that is within normal range, especially in the evening. Second, fever itself is part of the immune response, as discussed above. Pediatric guidelines generally advise treating fever for the child’s comfort rather than chasing a number on the thermometer. If a child has a temperature of 101 °F but is playing, drinking fluids, and reasonably happy, medication may not be necessary. Seizures and brain damage from fever are feared but appear to be associated with the rapidity of the temperature spike and with very high fevers above 107 °F, not with ordinary fevers in the 100–103 °F range that go untreated for a few hours.

Exercise, Heat, and Temperatures That Look Like Fever

Vigorous exercise can push core temperature to 38–39 °C (100.4–102.2 °F) without any pathology at all. The human body converts less than 20–30 percent of exercise energy into movement; the rest becomes heat. Healthy thermoregulation, primarily sweating and increased blood flow to the skin, keeps this manageable.²⁶ A runner who checks their temperature immediately after a hard workout and sees 101 °F is probably fine. The concern arises when core temperature exceeds about 40 °C (104 °F) and is accompanied by confusion, nausea, or collapse, which signals exertional heat stroke and requires immediate cooling.

Medications That Shift Your Thermostat

Certain common medications can interfere with your body’s ability to regulate temperature, raising the risk of overheating during hot weather or exercise. A systematic review and meta-analysis found moderate evidence that drugs with strong anticholinergic properties raised core temperature by about 0.4 °C under heat stress, while non-selective beta-blockers and anti-Parkinson’s medications also produced small but measurable increases.²⁷ Diuretics, especially when combined with blood-pressure medications, can compound the problem by depleting fluids and electrolytes. Psychotropic drugs and medications that reduce sweating or impair thirst recognition add further risk.²⁸

Elderly people taking multiple medications are at the highest risk for heat-related illness, particularly during heat waves. If you take any of these drug classes and spend time outdoors in the heat, extra hydration and more frequent breaks from sun exposure are worth the effort. This is also one reason why heat-wave mortality disproportionately affects older populations: the combination of an aging thermoregulatory system, lower baseline temperatures, and medication effects stacks the deck.