The famous 98.6 °F (37.0 °C) benchmark is no longer considered an accurate average for human body temperature. Multiple large studies show that the true average for healthy adults today runs closer to 97.5–97.9 °F, and it has been trending downward for over a century. The old number traces back to a single nineteenth-century physician’s measurements, and modern researchers have found enough reason to question it that some clinicians argue fever thresholds should shift downward too.
Where 98.6 Came From
The number originated with Carl Wunderlich, a German physician who published his work on body temperature in the 1860s. Wunderlich collected over a million readings from roughly 25,000 patients and declared 37.0 °C (98.6 °F) the standard normal value. His contribution was enormous: before his era, medical thermometry was rudimentary, and it took centuries of slow progress after Galileo’s thermoscope for temperature measurement to become a routine clinical tool.1ScienceDirect (Infrared Physics & Technology). The historical development of temperature measurement in medicine But Wunderlich was working with mercury thermometers of his time, measuring axillary (armpit) temperatures, and studying a population with far higher rates of chronic infection than we see in wealthy nations today. His figure was remarkably useful for over a century, but it was always a snapshot of a specific population at a specific moment in history.
The Temperature Drop Over Time
A landmark study published in eLife analyzed body temperature records from three large cohorts spanning birth years from the early 1800s through the late 1990s. The trend was unmistakable: body temperature declined steadily across the entire period, dropping about 0.03 °C per birth decade in both men and women. Over roughly two centuries, that adds up to a decline of about 0.59 °C in men.2eLife. Decreasing human body temperature in the United States since the Industrial Revolution The pattern held across racial groups, suggesting a broad biological shift rather than a quirk of one demographic.
This finding could be dismissed as a peculiarity of industrialized Western populations if it had not been replicated elsewhere. Among the Tsimane, an indigenous group living in the Bolivian Amazon with a very different lifestyle and disease environment, researchers recorded nearly 18,000 temperature measurements from over 5,400 adults and teenagers between 2002 and 2018. Average body temperature among the Tsimane dropped by about 0.9 °F in just 16 years, a pace even faster than what was seen in the U.S. data.3PubMed Central. Rapidly declining body temperature in a tropical human population The fact that a tropical, subsistence-level population showed the same directional shift argues against explanations that depend solely on air conditioning, sedentary lifestyles, or other hallmarks of modern industrial life.
Why Humans Are Running Cooler
The leading explanation is a dramatic reduction in chronic inflammatory burden. In Wunderlich’s era, tuberculosis, syphilis, periodontal disease, and untreated wound infections were far more prevalent. Chronic inflammation elevates baseline body temperature, so a population swimming in low-grade infection would naturally register warmer. As sanitation, antibiotics, and vaccines slashed the rate of persistent infections over the twentieth century, overall inflammatory load dropped, and baseline temperatures came down with it.3PubMed Central. Rapidly declining body temperature in a tropical human population
Reduced basal metabolic rate is another plausible contributor. Physical labor has decreased, climate-controlled environments reduce the metabolic cost of thermoregulation, and nutritional changes may have shifted how much heat the body generates at rest. The Tsimane data complicate this story somewhat, because the Tsimane remain physically active and lack widespread climate control, yet their temperatures dropped anyway. That has led researchers to suspect that even modest improvements in infection treatment and public health may be enough to produce the trend, regardless of activity level or housing.
A less familiar piece of the puzzle involves the gut microbiome. The trillions of microbes living in the intestines are not passive residents; they produce heat through anaerobic metabolism. Animal research has shown that surgically removing a major portion of gut microbial biomass can reduce total body heat production by roughly 8%, an outsized contribution given that the microbial mass involved represents only about 1–2% of body weight.4Journal of Experimental Biology. Gut microbial contributions to thermogenesis If modern dietary patterns and antibiotic use have shifted the composition or total mass of gut microbes, that could subtly influence baseline temperature across whole populations. This line of research is still young, but it offers a concrete biological pathway that connects changes in human lifestyle to measurable shifts in internal heat production.
Your Temperature Changes Throughout the Day
Even if 98.6 °F were correct as a population average, it would be misleading as a personal reference because body temperature is never static. It follows a strong daily rhythm, bottoming out in the early morning hours (typically between 6:00 and 8:00 a.m.) and peaking in the late afternoon or early evening (between 6:00 and 8:00 p.m.).5PubMed. The daily, weekly, and seasonal cycles of body temperature analyzed at large scale The swing can be around 1 °F or more, which means a temperature that looks “normal” at 7 a.m. might be genuinely low for that same person at 5 p.m., and vice versa.
Seasonal and even weekly rhythms exist too. The daily low point tracks sunrise timing across the year, consistent with the role of sunlight as a signal for the body’s internal clock.5PubMed. The daily, weekly, and seasonal cycles of body temperature analyzed at large scale The practical upshot is that a single reading in a doctor’s office tells you relatively little without knowing when it was taken and what that person’s individual baseline looks like at that time of day.
Age, Sex, and Other Individual Factors
Older adults tend to run cooler. A study of nursing home residents found a mean oral baseline of about 97.4 °F, well below the traditional standard.6PubMed. Fever response in elderly nursing home residents: are the older truly colder? This has real clinical consequences, because fever thresholds defined by absolute numbers can miss infections in older patients. In that same study, nearly half of infection episodes produced a “blunted” fever response in which the maximum temperature never reached 101 °F. Some of those patients still had a meaningful temperature rise from their personal baseline, but because the starting point was low, the final number did not cross the threshold that would trigger concern. For caregivers, this means that a reading of, say, 99.5 °F in a frail 85-year-old can represent a more significant change than 101 °F in a younger person.
Sex and hormonal status also shift the numbers. Women generally have slightly higher core temperatures than men, with one controlled study finding a difference of about 0.23 °C.7PubMed Central. Core body temperature in obesity On top of that baseline difference, the menstrual cycle produces a well-documented biphasic pattern. In the first half of the cycle, when estrogen is dominant, core temperature is lower. After ovulation, rising progesterone pushes it up by roughly 0.3 to 0.7 °C.8PubMed Central. Temperature regulation in women: Effects of the menstrual cycle This shift is robust enough that tracking basal body temperature has been used for decades as a fertility-awareness method, and newer wearable devices that measure wrist skin temperature confirm the same biphasic pattern, with the lowest readings coinciding with estrogen and LH peaks around ovulation and the highest readings occurring during the luteal phase.9npj Women’s Health. Understanding wrist skin temperature changes to hormone variations across the menstrual cycle
Body composition, which many people assume plays a major role, turns out to be less important for core temperature than expected. A study comparing obese and non-obese subjects found no significant difference in mean daily core temperature between the two groups.7PubMed Central. Core body temperature in obesity Fat tissue does affect skin temperature in complex ways; for example, higher body fat is associated with warmer fingertips but cooler abdominal skin, likely because of how fat insulates underlying tissue differently across body regions.10The American Journal of Clinical Nutrition. Adiposity and human regional body temperature But when it comes to the deep core temperature that a standard thermometer aims to estimate, being heavier does not systematically make you warmer or cooler.
Where You Measure Matters More Than People Realize
The 98.6 °F figure was originally derived from axillary measurements, and modern readings can come from the mouth, ear, forehead, or rectum. These sites do not agree with each other particularly well. In emergency department comparisons, roughly a third of ear and temporal artery (forehead) readings differed from rectal temperature by 0.5 °C or more, and half of oral readings showed that level of discrepancy.11Emergency Medicine Journal. Temperature measurement in the adult emergency department: oral, tympanic membrane and temporal artery temperatures versus rectal temperature Ear measurements were the most consistent of the non-rectal options, but even those had a standard deviation from rectal temperature of 0.4 °C.
In pediatric settings, similar variation appears: non-contact forehead thermometers and temporal artery devices tend to read lower than oral, while some temporal artery devices can read higher.12PubMed Central. What is the Difference Between the Different Types of Thermometers? This variability means that comparing a forehead reading to an oral-temperature chart, or vice versa, introduces a layer of imprecision on top of whatever biological variation already exists. The kind of thermometer you use can shift your apparent temperature by half a degree or more in either direction.
Even the same type of thermometer introduces variability depending on its technology. Early comparisons of digital and mercury clinical thermometers found that while average accuracy was similar, electronic thermometers showed much more fluctuation from one measurement to the next. Repeated readings with a digital thermometer differed by 0.5 °C or more between 9 and 23% of the time, compared with under 1% for mercury thermometers.13PubMed Central. A comparison of mercury and digital clinical thermometers Mercury thermometers are now largely phased out for safety reasons, so we live with the slightly noisier digital alternative. That is worth keeping in mind if you ever get a reading that seems unexpectedly high or low: retaking it is reasonable before jumping to conclusions.
What This Means for Defining Fever
If normal body temperature has drifted downward, the standard fever threshold may need to follow. Most clinical guidelines still define fever as a temperature above 38.0 °C (100.4 °F), a cutoff inherited from the era when 37.0 °C was the assumed baseline.14PubMed Central. Threshold for defining fever varies with age, especially in children: A multi-site diagnostic accuracy study Researchers have started to push back against this rigid standard, arguing that clinging to 38.0 °C as a universal threshold may lower sensitivity for detecting illness, particularly in populations whose baselines are well below the historical average.15Open Forum Infectious Diseases. Fever and Fever of Unknown Origin: Review, Recent Advances, and Lingering Dogma
The issue is especially acute in older adults, whose low baselines mean a genuinely significant rise might only reach 99 or 100 °F, numbers that most triage protocols would dismiss as unremarkable. But the problem extends to younger adults too. If your personal baseline hovers around 97.5 °F, reaching 100.0 °F represents a 2.5-degree jump, which is physiologically comparable to someone with a 98.6 baseline hitting 101.1. Both people are mounting similar immune responses, but only the second one gets flagged by the traditional fever definition.
This concern is more than theoretical. In sepsis, the body’s temperature response follows a bimodal pattern: patients tend to develop either fever or hypothermia, with normothermia being uncommon. Both extremes, very high fever and abnormally low temperature, are associated with worse outcomes.16PubMed Central. Fever and hypothermia represent two populations of sepsis patients and are associated with outside temperature A rigid cutoff that defines “fever” only above a certain point may cause clinicians to overlook the subset of patients whose bodies respond to severe infection by dropping temperature instead of raising it.
Wearable Devices and the Case for Personal Baselines
Continuous temperature monitoring through smartwatches and wristband devices is starting to challenge the whole concept of a single normal number. A study using wearable ring sensors found substantial variation between individuals in both their average finger skin temperature and their daily range. When these same people developed symptomatic illness, their temperatures rose by an average of about 0.63 °C compared to their own baselines, but the spread was wide enough that a single fixed threshold like 38 °C would not have reliably caught all of them.17Scientific Reports. Feasibility of continuous fever monitoring using wearable devices
The implication is that knowing your personal baseline and flagging deviations from it may be more clinically useful than comparing any single reading to a population-wide cutoff. This is already how fertility tracking works: the shift of a few tenths of a degree after ovulation only means something relative to the individual’s own follicular-phase readings, tracked consistently over time.18PubMed. Descriptive analysis of the relationship between progesterone and basal body temperature across the menstrual cycle The same logic applies to detecting early fever. A wearable that knows your typical nighttime temperature can flag a 0.5-degree deviation before you even feel symptomatic, something a spot-check at the doctor’s office cannot do.
We are still in the early days of this shift. Wrist and finger skin temperatures are not the same as core body temperature, the algorithms translating one to the other are imperfect, and most clinicians still rely on point-in-time readings with traditional devices. But the direction is clear: the era of comparing everyone to a single fixed number is fading, and individualized baselines are becoming the more useful frame of reference.
Exercise and Heat Stress
It is worth noting that perfectly healthy people routinely exceed what would traditionally be called “fever” during exercise, especially in warm environments. Prolonged physical activity naturally drives core temperature upward as muscles generate heat faster than the body can dissipate it through sweating and skin blood flow.19PubMed. Exercise under heat stress: thermoregulation, hydration, performance implications, and mitigation strategies Marathon runners and military recruits in hot-weather training regularly hit core temperatures above 39 °C (102.2 °F) without being sick. This is one of several reasons that context matters more than any absolute number: the same reading can mean “running a race” or “fighting an infection,” and those are obviously very different situations.
Higher BMI does affect how the body handles heat during exercise or surgery, even though it does not change resting core temperature. In surgical patients, those with higher BMIs showed delayed onset of vasoconstriction and maintained slightly higher tympanic temperatures during the early minutes of anesthesia compared to leaner patients.20PubMed Central. The effect of body mass index on perioperative thermoregulation The insulating effect of fat changes the dynamics of heat loss rather than heat production, which is why core temperatures at rest look similar across body types but diverge under conditions that stress the thermoregulatory system.
The Children Question
Parents tend to be particularly anxious about temperature numbers, and the pediatric guidelines reflect an awareness that a single threshold does not fit all ages. The standard definition of fever in children remains above 38.0 °C (100.4 °F), with rectal thermometers recommended for children under three years old because other methods are less reliable in young kids.14PubMed Central. Threshold for defining fever varies with age, especially in children: A multi-site diagnostic accuracy study But researchers have argued that this fixed threshold needs reexamination, because children’s baseline temperatures and fever responses vary by age group. An infant’s temperature regulation is less mature than a school-age child’s, and the clinical significance of a given reading depends partly on the child’s developmental stage.
What matters more than hitting a specific number is the pattern: how quickly the temperature is rising, how the child looks and behaves, and whether the temperature responds to the body’s own regulatory efforts or to fever-reducing medication. A child at 100.2 °F who is listless and refusing fluids is more worrying than one at 101.5 °F who is still playing and drinking normally. The thermometer gives you a data point, but it is never the whole story, and anchoring too firmly on 98.6 as “normal” or 100.4 as “fever” can distort how parents and even some clinicians interpret the reading.