A reading of 37.8°C (100.0°F) sits right on the clinical fence. The Merck Manual defines oral fever as anything above 37.8°C, while Harrison’s Principles of Internal Medicine considers a late-afternoon oral reading of 37.7°C the upper edge of normal. Whether 37.8 counts as a fever depends on where in your body you took the reading, what time of day it is, how old you are, and which clinical definition your doctor uses. The short version: 37.8 measured rectally is almost certainly normal, while 37.8 measured under the arm probably is a fever.
The 37.0°C Benchmark Was Never Quite Right
The number most people treat as “normal” body temperature, 37.0°C or 98.6°F, comes from a German physician named Carl Reinhold August Wunderlich, who published it in the 1860s based on millions of readings taken with the thermometers of his era. A landmark 1992 study re-examined this figure using modern instruments and found that average oral temperature was closer to 36.8°C (98.2°F), and that the upper boundary of the normal range was about 37.7°C (99.9°F) rather than the traditionally cited 38.0°C (100.4°F).1PubMed. A critical appraisal of 98.6 degrees F, the upper limit of the normal body temperature, and other legacies of Carl Reinhold August Wunderlich
More recent evidence suggests that human body temperature has been drifting downward for the past two centuries. A large analysis spanning nearly 200 years of birth cohorts found a steady decline of roughly 0.03°C per decade in both men and women.2eLife. Decreasing human body temperature in the United States since the Industrial Revolution Some current outpatient data put the typical normal range between 36.2°C and 36.8°C.3Medical Hypotheses. Climate change and the decline in human core body temperature: A hypothesis of physiological adaptation If true normal is now lower than Wunderlich’s figure, then 37.8°C becomes relatively more significant. A reading that once looked borderline might now represent a more meaningful departure from your actual baseline.
Where You Measure Changes the Number
Body temperature is not uniform. Your core, deep inside the trunk, runs warmer than your skin, your mouth, or your armpit. Every measurement site is a proxy for core temperature, and each proxy introduces its own offset. That offset is why 37.8 can mean different things depending on how the reading was taken.
A systematic review of studies with strong evidence found that normal temperature ranges differ substantially by site: oral temperatures ranged from 33.2°C to 38.2°C, rectal from 34.4°C to 37.8°C, tympanic (ear) from 35.4°C to 37.8°C, and axillary (armpit) from 35.5°C to 37.0°C.4PubMed. Normal oral, rectal, tympanic and axillary body temperature in adult men and women: a systematic literature review These are wide ranges, but the key pattern is consistent: rectal reads highest, oral and ear readings cluster together in the middle, and armpit reads lowest.
In children presenting to the emergency department, temporal artery readings ran about 0.2°C below rectal, while axillary readings were a full 0.9°C below rectal.5The American Journal of Emergency Medicine. Temporal artery and axillary thermometry comparison with rectal thermometry in children presenting to the ED A study in children with Kawasaki disease found that an axillary reading of 37.5°C was the best threshold for catching an oral fever of 38.0°C or above, with sensitivity and specificity both around 94%.6PubMed Central. Axillary, Oral, and Rectal Routes of Temperature Measurement During Treatment of Acute Kawasaki Disease
Here is what those offsets mean for interpreting 37.8°C:
- Rectal: 37.8°C is within the upper end of normal. Rectal fever thresholds usually start at 38.0°C or 38.2°C.
- Oral: 37.8°C is borderline. Some guidelines call it the start of fever; others place the line just above it.
- Axillary: 37.8°C is well above the normal armpit range (which tops out around 37.0°C in most studies) and almost certainly indicates fever.
- Tympanic: 37.8°C sits at the upper limit of normal ear temperature, so it could go either way depending on the person and the time of day.
Competing Clinical Definitions
There is no single, universally agreed-upon number that defines fever. A review of major medical references found that Harrison’s Principles of Internal Medicine defines fever as a core (rectal) temperature between 37.5°C and 38.3°C, a skin (axillary) temperature above 37.2°C, a morning oral temperature above 37.2°C, or a late-afternoon oral temperature above 37.7°C. The Merck Manual, meanwhile, sets the oral fever line at above 37.8°C and the rectal line at above 38.2°C.7PubMed Central. Defining Fever Those two widely used references disagree by a few tenths of a degree, and they also note that lower thresholds should be used for frail elderly patients.
So by Harrison’s late-afternoon oral definition, 37.8°C is one tick above the 37.7°C ceiling and counts as a fever. By the Merck Manual’s oral definition, 37.8°C is the threshold itself. If you measured rectally, neither reference would call 37.8 a fever. The definition you land in depends on the thermometer you picked up and which textbook your clinician trained on.
Time of Day Matters More Than You Might Think
Your core body temperature follows a roughly sinusoidal daily rhythm, bottoming out between about 3:00 and 5:00 a.m. and peaking between 3:00 and 5:00 p.m.8Polish Hyperbaric Research. The circadian rhythm of core body temperature (Part I): The use of modern telemetry systems to monitor core body temperature variability In healthy people, this swing amounts to roughly half a degree Celsius. That is a big deal for interpreting a borderline number. A reading of 37.8°C at 6:00 a.m. is more worrisome than the same reading at 4:00 p.m., because it represents a bigger departure from where your temperature would normally be at that hour.
Harrison’s definition implicitly acknowledges this by setting different oral thresholds for morning (above 37.2°C) and late afternoon (above 37.7°C). If you take your temperature at the wrong time and don’t account for the daily cycle, you can either miss a fever or falsely flag one.
Age, Sex, and the Menstrual Cycle
Your personal baseline is shaped by who you are, not just when and where you measure. Age is one of the biggest modifiers. Research on hospitalized patients with infections found that for each decade of age, the average fever temperature during the first three days of illness dropped by about 0.15°C.9The American Journal of the Medical Sciences. The Relationship between Age and Fever Magnitude Older adults not only run cooler at baseline but also mount lower peak temperatures when they are sick. A reading of 37.8°C in an 80-year-old with a suspected infection may represent a more serious fever response than the same number in a 25-year-old. Interestingly, though, experimental work with endotoxin has shown that older adults can have a more prolonged fever response even when peak temperatures are lower.10PubMed Central. Ageing is associated with a prolonged fever response in human endotoxemia
Sex matters too. The systematic review of normal ranges noted that oral temperatures for women spanned 33.2°C to 38.1°C, compared with 35.7°C to 37.7°C for men.4PubMed. Normal oral, rectal, tympanic and axillary body temperature in adult men and women: a systematic literature review Part of this wider range in women traces to the menstrual cycle. Core temperature rises 0.3°C to 0.7°C during the luteal phase (the roughly two weeks after ovulation) compared with the follicular phase, driven by progesterone.11PubMed Central. Temperature regulation in women: Effects of the menstrual cycle For a woman in the luteal phase, a reading of 37.8°C might just be the top of her normal hormonal bump, while the same reading in the follicular phase would be more unusual.
Common Artifacts That Throw Off a Reading
If you drink something cold or hot shortly before placing an oral thermometer, you are not measuring your body temperature at all. You are measuring something closer to the temperature of what you drank. A study of the effect of iced water on oral readings found an average initial drop of 1.3°C, with some individuals seeing drops as large as 4.6°C. The time needed for temperatures to return to baseline depended on age: about 15 minutes for people under 40, 20 minutes for those between 40 and 59, and 30 minutes for people 60 and older.12PubMed. Temperature lowering after iced water. Enhanced effects in the elderly
Hot beverages push readings in the other direction. One study found an immediate mean elevation of about 1.4°C after a hot drink, which would have disqualified seven out of ten blood-donation candidates. Readings returned to acceptable levels within five minutes in most cases. Cold beverages took longer to clear, with some subjects still below baseline at 10 minutes and a few not recovering until 30 minutes out.13PubMed. The effect of hot beverages, cold beverages, and chewing gum on oral temperature A separate study found that hot and cold liquids significantly influenced oral temperature for about seven to nine minutes after ingestion, while ear thermometer readings were unaffected.14The American Journal of Emergency Medicine. A comparison of oral, rectal, and tympanic membrane-derived temperature changes after ingestion of liquids and smoking
Ear thermometers have their own pitfalls. Earwax buildup tends to lower the reading, with one study in older adults finding drops of 0.3°C or more in nearly half of subjects with impacted cerumen.15PubMed. Effect of cerumen on infrared ear temperature measurement Ear infections can also alter the local temperature of the eardrum and throw off results.16PubMed Central. Accuracy of tympanic temperature measurement using an infrared tympanic membrane thermometer On the reassuring side, at least one study found that how deeply you insert the ear probe doesn’t seem to significantly affect the reading.17Swiss Medical Weekly. Limitations of infrared ear temperature measurement in clinical practice
Forehead thermometers, popular since the pandemic, are the most sensitive to environmental conditions. A study comparing forehead, armpit, and oral temperatures found that forehead readings averaged about 0.5°C lower than oral and armpit readings and correlated poorly with them, partly because ambient temperature influenced the forehead measurement.18PubMed Central. A Comparative Study of Forehead Temperature and Core Body Temperature under Varying Ambient Temperature Conditions If you have been standing outside on a cold day, a forehead scan of 37.8 is quite different from the same reading in a warm clinic.
How Fever Actually Works
Fever and simply being hot are not the same process. In fever, your brain’s thermostat, the hypothalamus, deliberately raises its set point in response to signaling molecules released during infection or inflammation. Your body then actively works to reach that new set point through shivering, vasoconstriction, and other heat-conserving mechanisms.19PubMed Central. Fever in Sepsis Revisited: Is a Little Heat What We Need? In hyperthermia, by contrast, the thermostat stays where it is but the body’s cooling systems get overwhelmed, as in heat stroke or certain drug reactions.20DeckerMed Medicine. Hyperthermia, Fever, and Fever of Undetermined Origin
This distinction matters because fever is not a malfunction. It is a deliberately orchestrated immune response that has been conserved in warm-blooded and cold-blooded vertebrates for over 600 million years of evolution, and it has been shown to confer a survival benefit during infection.21PubMed Central. Fever and the thermal regulation of immunity: the immune system feels the heat A temperature of 37.8 caused by a viral infection is your immune system doing something purposeful. The same temperature caused by sitting in a hot car with no ventilation is a danger sign that requires cooling.
Temporal Artery Thermometers and the Sensitivity Problem
Temporal artery thermometers, the forehead-swipe type, have become standard in many clinics and hospitals. They are fast and noninvasive, which is why they spread rapidly during mass screening efforts. But their ability to detect a genuine fever is less reliable than many people assume. A systematic review in adults found that while their specificity was acceptable, their sensitivity was low, meaning they are better at confirming you don’t have a fever than at catching one you do have.22PubMed. Agreement of infrared temporal artery thermometry with other thermometry methods in adults: systematic review
A large meta-analysis quantified the problem. Across more than 9,800 readings, temporal artery thermometers had a pooled sensitivity of only about 59% for fever detection. Performance was better in children (around 77% sensitivity) than in adults (around 48%).23PubMed Central. The impact of age on comparative diagnostic accuracy of temporal artery thermometers and non-contact infrared thermometers for fever detection: a systematic review and meta-analysis That means a temporal artery scan could easily read 37.8°C in someone whose rectal temperature is 38.5°C or higher. In a comparison at an African outpatient clinic, the temporal artery thermometer actually came out on top among three infrared devices, with 88% sensitivity and 88% specificity, though this was in a setting specifically optimized for the comparison.24PubMed Central. Are all thermometers equal? A study of three infrared thermometers to detect fever in an African outpatient clinic
The practical takeaway: if your temporal artery thermometer reads 37.8, and you feel genuinely unwell, it is worth confirming with an oral or rectal thermometer before deciding whether or not you have a fever.
Mass Screening and Its Limits
During pandemic-era entry screening at airports, offices, and hospitals, thermal cameras and handheld infrared scanners became the first line of defense. These tools were designed to flag anyone with an elevated surface temperature. Research on infrared thermal detection systems found that the best-performing devices could reach roughly 90% sensitivity and 80–86% specificity for identifying fever under controlled conditions.25PubMed Central. Comparison of 3 Infrared Thermal Detection Systems and Self-Report for Mass Fever Screening But real-world performance varied. A study in a tropical healthcare setting found that a handheld infrared thermoscope caught only about 29% of confirmed febrile patients, while self-reported fevers actually had better sensitivity at 88%.26PubMed Central. Comparison of Infrared Thermal Detection Systems for mass fever screening in a tropical healthcare setting
The gap between lab-tested performance and field performance comes down to exactly the variables discussed throughout this article: ambient temperature, recent food or drink, time of day, individual variation in baseline. A person whose normal resting temperature is 36.2°C might register 37.8°C on a screening scan and actually be running a meaningful fever, while someone whose baseline runs higher could sail through the scanner at the same reading without actually being sick. Mass screening is useful for catching the obvious cases but fundamentally cannot resolve borderline readings like 37.8.
Wearable Devices and the Shift Toward Personal Baselines
The biggest conceptual shift in fever detection is moving away from a single universal threshold and toward tracking each person’s individual pattern. A wearable patch tested in a clinical trial measured body temperature continuously and was able to detect clinic-confirmed fevers with about 90% sensitivity, often flagging a fever a median of four hours before the clinical thermometer caught it.27PubMed Central. A Novel Wearable Device for Continuous Temperature Monitoring & Fever Detection A pilot study of two wearable devices in cancer patients at risk for dangerous temperature spikes found that both correlated reasonably well with clinical tympanic readings and showed potential for early detection.28PubMed Central. Continuous temperature monitoring for the detection of fever in haemato-oncology patients: a pilot study of two wearable devices (THERMAL)
Research using consumer wrist-worn wearables has reinforced a key finding: there is substantial variation between individuals in both mean temperature and daily range. While average wrist-measured temperature in one study was lower than core temperature (around 31.2°C, since the wrist is skin surface), the researchers found a significant rise during illness and concluded that using a single fixed threshold like 38°C across all people would not be appropriate.29Scientific Reports. Feasibility of continuous fever monitoring using wearable devices The future of fever detection likely looks less like “is this number above 38?” and more like “is this number elevated compared to what this particular person’s body normally does at this time of day?” Under that framework, 37.8°C would be a fever for some people and perfectly normal for others, and the device would know the difference.
Neonates and Young Children Are a Special Case
For babies, fever thresholds and measurement approaches differ from adults in ways that catch many new parents off guard. Rectal temperature remains the reference standard in very young infants, and a reading of 38.0°C or above rectally in a baby under three months is treated as a medical urgency regardless of how well the baby looks. Interestingly, one study found that in neonates (birth to three months), an infrared ear thermometer using age-adjusted fever thresholds outperformed a rectal thermometer using fixed thresholds for predicting illness, with 93.5% accuracy versus 72.8%.30PubMed Central. Threshold for defining fever varies with age, especially in children: A multi‐site diagnostic accuracy study The improvement came from adjusting the threshold to the baby’s age rather than applying one number to all patients.
Temporal artery thermometers, meanwhile, have a different relationship with rectal readings in neonates than they do in adults. In one study of critically ill newborns, temporal artery readings averaged 0.43°C warmer than rectal readings, the reverse of the pattern typically seen in older children and adults.31PubMed. Comparison of Temporal Artery and Rectal Temperature Measurement During Cooling and Rewarming in Neonates Treated for Hypoxic Ischemic Encephalopathy These neonates were being treated for a specific brain injury, so the results may not generalize to healthy babies, but they illustrate how unpredictable the offsets between measurement sites can be in the youngest patients. For infants, the answer to “is 37.8 a fever?” depends even more strongly on how the temperature was obtained and on the child’s age in weeks.