Average body temperature is dropping

A few days ago, I started feeling a little lousy, ‘under the weather’ as they say in the UK, and decided to take my temperature and sure enough I had a low fever of about 100.6F. In these pandemic days, even though I am vaccinated and take precautions, wear masks in public places and as much as possible only associate with vaccinated people, there is always the chance of breakthrough infections so even though I had none of the other symptoms of covid such as loss of taste and smell, I decided to take a home test to see if I had contracted covid. The test came beck negative, which was a relief, but these rapid tests are not that reliable so one can never be sure, so I decided to self-isolate until my fever went back to normal.

And this was the. problem: what I was not sure what the ‘normal’ temperature is. The conventional normal is 98.6F (37C) but I knew that my body temperature when I am not sick is much lower, around 97F. So if my temperature dropped to around 98.6F, which it did after about 12 hours, was my fever gone? Or did I still have a low-grade fever?

It turns out that the conventional measure of what constitutes normal body temperature is way overdue for a re-evaluation. For one thing, normal temperatures vary quite a bit from individual and even for any given individual, their body temperature varies with many factors, including the time of day, with temperatures being higher later in the day. So what is normal in the evening might not be so the next morning. Furthermore “Women tend to have higher body temperature than men, and younger people tend to have higher temperatures than older folks.”

But what I found most interesting is that average body temperatures have been dropping over time.

An analysis of 20 studies between 1935 and 1999 found that the average oral temperature was 97.5˚ F. And a 2017 study of more than 35,000 people found an average body temperature of 97.9˚ F.

On this last point, a remarkable new study is among the best to make a case that normal body temperature has been drifting down over the last two centuries.

In this study, researchers analyzed temperature recordings from three periods of time over 157 years:

  • 1860–1940: A mix of armpit and oral temperatures of nearly 24,000 veterans of the Civil War were measured.
  • 1971–1975: Oral temperatures of more than 15,000 people from a large population study (the National Health and Nutrition Examination Survey) were analyzed.
  • 2007–2017: Oral temperatures of more than 150,000 people in another large research project (the Stanford Translational Research Integrated Database Environment) were reviewed.

During the nearly 160 years covered by the analysis, the average oral temperature gradually fell by more than one degree. As a result, the “new normal” seems closer to 97.5˚ F.

This observation held up even after accounting for age, gender, body size, and time of day.

So what might be the cause for this drop? It turns out that changes in the way body temperature is measured is not likely to be a factor. More likely is a change in metabolic rates, which is one of the biggest determinants of body temperature and a “lower metabolic rate in modern times could be due to higher body mass (some studies link this with lower metabolic rate), or better medical treatments, preventive measures, and overall health.”

Having a better measure for what constitutes normal body temperature is important because it is a key measure of one’s health. Physicians might miss early warning signs of illness if they think a patient’s temperature of around 99.5F is close enough to normal to not worry about, when. in reality that person’s normal temperature may be 97F or less.

So what to do?

Fever is typically any temperature above 100˚ F. The most common cause of fever is any infection in the body, but there are other causes, including heat stroke or a drug reaction. Although you can be sick with a normal temperature, body temperature is clearly an important and useful indicator of health.

While news that the normal body temperature may be drifting down over time is intriguing, it is not cause for alarm — and it doesn’t mean the definition of fever should change. We’ll need to rely on additional research to tell us how important these findings may be. In the meantime, it’s probably time to abandon the assumption that 98.6˚ is a normal temperature. Something closer to 97.5˚ may be more accurate.

After I felt back to ‘normal’, I took my temperature multiple times and it was around 96F in the morning, so clearly I am at the lower end of whatever can be considered the normal range of body temperatures.


  1. Katydid says

    At various points during the pandemic, I needed to take my temperature. Like you, I also learned that I am not typically 98.6, but in fact in the 97s. I’ve also wondered: if my usual temp is 97.5 and I measure a 99.5, does that mean I have a fever? If you call the doctor, you might be told you’re fine, but they base that on a “normal” temp of 98.6.

    Also, while the COVID vaxes I took raised my body temp slightly, the first shingles shot sent my temp into the 102 range for 3 days, also with fever, chills, body aches, headache, etc. Not looking forward to getting the next shot of that. I wondered just how dangerous a 102-point-something temp was to someone whose normal body temp is in the 97s.

  2. kenbakermn says

    My conjecture is that when the original study was done back in the 19th century (probably) the researchers used metric units and ended up with 37C, which is 98.6F. But maybe the true average was something closer to 36.5C or a bit higher, and it was rounded up to 37C to account for the imprecision of the measurement. Well, 36.5C is 97.7F, not too far off from current measurements.

  3. says

    I have no idea what t is in Freedumbheit, but in sensible units, my personal “normal” is in the interval 36,4-36,7 °C. At 37 °C, which I usually get around this time of year in the afternoon/early evening I feel generally unwell. I say that I feel sick enough to not be able to do anything but not so bad to warrant a doctor visit or to be able to stay in bed.

  4. marner says

    Before vaccinations anyway, most businesses were taking the temperature of every employee before allowing them to work. I have wondered if anyone has (will?) gathered all those millions upon millions of data points.

  5. Pierce R. Butler says

    … even after accounting for age, gender, body size, and time of day.

    Do these researchers (even have an agreed standard on) account(ing) for probable differences in amount of daily physical activity (i.e., labor) individually or by population? I would make a bet (except for having no way to get the answer) that my physiology ran a little warmer when I did lots of manual work daily.

  6. Bruce says

    I agree with Ken @#3.
    Also, the average temperature of about 37 C was published by a German in 1868. His thermometers were likely +/- 1C or 0.5 C at best, as were almost all ones used up through the 1975 studies, before digital ones became available.
    In other words, from the start in 1868 through the 1980s, “everyone”, knew that any temp was likely +/- 0.5 C 0.9 F at the very best. The fact that studies got a lot of measurements and averaged them and calculated the average to four digit precision doesn’t change the fact that all data that went in to every study from 1858-1975 or more was rounded off to the nearest half a Celsius or the nearest 1 degree F. So even if body metabolisms hadn’t changed, it was always pretentious to pretend to know the average to the nearest 0.1 degree.
    Then, we need to add in natural variations between people and metabolisms, etc.
    For all of human history until the last fifty years, the age rage person, by definition, was NOT overweight, but now we are (except Mano). So it is obvious that something poorly understood is different with bodies now vs 1975 or 1868. When Columbus sailed in 1492, he was likely not drinking high fructose corn syrup, etc.
    So, deviation from 37C or 98.6000 F is not surprising. The only surprising idea is that we still act as if we thought we should be comparing with a precise number.
    Wikipedia notes: … any oral temperature between 36.3 and 37.3 °C (97.3 and 99.1 °F) is likely to be normal.

  7. Dunc says

    His thermometers were likely +/- 1C or 0.5 C at best, as were almost all ones used up through the 1975 studies, before digital ones became available.

    Still pretty much the case -- your average consumer grade digitial thermometer may be precise to 0.1 degrees C, but if you read the specs, most of them only claim to be accurate to +/-1 degree C. If you want a thermometer that’s actually accurate to 0.1 C, you’ll need to pay for it, and to have it regularly recalibrated.

  8. anat says

    Body temperature is really dropping in ‘Western’ and high-income countries. See Decreasing human body temperature in the United States since the Industrial Revolution by Myroslava Protsiv, Catherine Ley, Joanna Lankester, Trevor Hastie, Julie Parsonnet from Stanford. The study accounts for differences in measurement technique/technology. Their proposed explanation:

    Although there are many factors that influence resting metabolic rate, change in the population-level of inflammation seems the most plausible explanation for the observed decrease in temperature over time. Economic development, improved standards of living and sanitation, decreased chronic infections from war injuries, improved dental hygiene, the waning of tuberculosis and malaria infections, and the dawn of the antibiotic age together are likely to have decreased chronic inflammation since the 19th century. For example, in the mid-19th century, 2–3% of the population would have been living with active tuberculosis (Tiemersma et al., 2011). This figure is consistent with the UAVCW Surgeons’ Certificates that reported 737 cases of active tuberculosis among 23,757 subjects (3.1%). That UAVCW veterans who reported either current tuberculosis or pneumonia had a higher temperature (0.19°C and 0.03°C respectively) than those without infectious conditions supports this theory (Supplementary file 1). Although we would have liked to have compared our modern results to those from a location with a continued high risk of chronic infection, we could identify no such database that included temperature measurements. However, a small study of healthy volunteers from Pakistan—a country with a continued high incidence of tuberculosis and other chronic infections—confirms temperatures more closely approximating the values reported by Wunderlich (mean, median and mode, respectively, of 36.89°C, 36.94°C, and 37°C) (Adhi et al., 2008).

  9. anat says

    To Bruce @8:

    Also, the average temperature of about 37 C was published by a German in 1868. His thermometers were likely +/- 1C or 0.5 C at best, as were almost all ones used up through the 1975 studies, before digital ones became available.

    Mercury thermometers had the ability to measure differences of 0.1C quite consistently. If I ran 37.0 or more in the morning I felt sick and was required to stay home. When I felt well I was 36.6 C in the morning. Evening temperature tended to be about 0.2C higher on typical days.

    recently, when I got the first dose of the Shingles vaccine I woke up feeling bad, my temperature was 37C. (And while my temperature was back to normal the following day, according to my Fitbit my resting heart rate kept rising and peaked 3 days post-vaccine.)

  10. anat says

    With the 3rd generation of the Oura ring it should be possible to track temperature constantly. That might yield interesting signals.

Leave a Reply

Your email address will not be published. Required fields are marked *