Why we get sick more in winter: cold, dry air, and indoor time

Start with a fact that sounds like a detail: in a 78-site comparison across the world, flu peaked in winter where the monthly air got cold and dry enough. Where it didn't, the peak tended to fall in the rainiest months instead (Tamerius et al., 2013). If cold were the whole story, warm places would simply have less flu. What they have is flu on a different calendar.

So "you catch colds in winter because it's cold" is a shortcut. This article walks through the mechanisms researchers have proposed, says what kind of evidence stands behind each one, and separates weather from behaviour where the data allow it. This is informational material, not medical advice. If symptoms are severe, getting worse, or not improving, that is a question for a doctor.

One more framing note. Almost nothing here is a trial in people. The evidence comes from population models, guinea pigs, mice, and human cells in a dish, and each section says which.

The pattern is real, and the tropics break it

A review of the field names two broad groups of contributors: changes in the environment (temperature, humidity) and changes in human behaviour (Moriyama et al., 2020). Temperature and humidity affect how long a virus stays viable and how well it spreads, and they also change how the immune system in your airway responds. The review covers both outdoor and indoor climate.

The 78-site study put numbers on the split. Where monthly average specific humidity or temperature fell below roughly 11 to 12 g/kg and 18 to 21 °C, flu tended to peak in the cold, dry season. Where they stayed above those levels, flu was more likely to peak in the months with the most rain, above 150 mm a month. In the tropics epidemics often came in the rainy season or showed no clear season (Tamerius et al., 2013).

That points away from "cold" as a single cause and toward a set of conditions that differ by climate. The rest of the article is about those conditions.

Dry air: absolute humidity, relative humidity, and your radiator

Two humidity numbers get mixed up constantly. Absolute humidity is how much water vapour is in the air. Relative humidity is that amount compared with what the air could hold at its current temperature. Cold air holds little water, so winter air outdoors is low in absolute humidity. And when you heat that air indoors, the amount of vapour stays the same while the air's capacity grows, so relative humidity drops. That is ordinary physics, not a finding from the studies below.

Which of the two matters for flu is not settled, and the studies used different ones.

A US analysis found that the start of the wintertime rise in flu-related mortality followed unusually low absolute humidity in the preceding weeks. A model in which absolute humidity alone modulates how transmissible the virus is reproduced the seasonal cycle, and the authors wrote that this modulation "is sufficient to produce this observed seasonality" (Shaman et al., 2010). Note what kind of result this is: an association plus a model, applied to temperate regions. It shows the idea is enough to explain the curve. It does not show that it is the only thing producing it.

The animal work is more direct. In 20 experiments with guinea pigs, run at relative humidity from 20 to 80% and temperatures of 5, 20, or 30 °C, both cold and dry conditions favoured transmission through the air. At 5 °C the animals shed virus at peak levels for about 40 hours longer than at 20 °C (Lowen et al., 2007). That speaks to the virus and its spread. It was done in a lab, in guinea pigs.

A second line concerns the host. Mice kept in low relative humidity got more severe disease after being exposed to influenza. Breathing dry air impaired the airway's mucus-clearing machinery, the innate antiviral defence, and tissue repair (Kudo et al., 2019). The paper discloses that the work was supported in part by a gift from the Condair Group, a humidification company. Again: mice, not people.

So dry air has support from three directions (a population model, guinea pig transmission, mouse immunity), and each direction is indirect for a person in a heated room. None of the studies measured indoor humidity in homes and followed who got sick.

A cooler nose

The second idea is older. In 2002 Eccles noted that most explanations for cold-season infections amounted to either "no explanation" or "winter crowding", and proposed another: breathing cold air cools the lining of the nose enough to inhibit local defences such as mucociliary clearance and the activity of white cells (Eccles, 2002). It was a hypothesis. Later lab work has partly backed it at the level of cells.

One piece of that work comes from the common cold virus. Most human rhinovirus isolates replicate better at nasal-cavity temperatures (33 to 35 °C) than at the core body temperature of 37 °C. In mouse airway cells, the interferon response, a key antiviral signal, was weaker at 33 °C, and the authors concluded the virus replicates preferentially at nasal temperature "due, in part, to a less efficient antiviral defense response of infected cells at cool temperature" (Foxman et al., 2015). Those are mouse cells in a dish.

Another is closer to human biology. When stimulated, human nasal lining releases a swarm of extracellular vesicles, tiny particles that bind and neutralize viruses. In human nasal epithelial cells and surgical nasal tissue, cold exposure impaired this: fewer vesicles were released, less microRNA was packed into them, and their antiviral binding was weaker. The authors describe it as a quantitative, mechanistic explanation for seasonal variation (Huang et al., 2023). It is human material, but it is still tissue and cells, with no population data attached.

There is a loose end. The guinea pig study also compared innate immune responses at 5 °C and 20 °C and found them comparable (Lowen et al., 2007). In that system, cold did not weaken innate immunity. Mouse cells and human nasal tissue point one way, guinea pigs another. The species differ, the viruses differ, the tissues differ, and nobody has resolved the disagreement. Read the cold-nose mechanism as plausible and supported at the cell level, and as unsettled for the whole person.

People indoors, and close together

Not everything is weather. School holidays are an unusually clean natural experiment, because they change who meets whom without changing the sky. French surveillance data showed that holidays cut the rate at which influenza passed to children by 20 to 29%, with no detectable change in adults' contact patterns. The authors estimate that holidays averted 16 to 18% of seasonal flu cases overall, and 18 to 21% in children (Cauchemez et al., 2008).

A caution about reading that. This is flu, in France, and it is about schools. It says that where people gather matters to the size of an epidemic. It cannot tell you how much of a given person's winter colds comes from the weather and how much from the commute and the office. None of these sources gives a split for an individual, and the review literature counts both environment and behaviour as contributors (Moriyama et al., 2020).

The mechanisms may also feed each other. People spend more time indoors when it is cold, and indoor air in a heated room is dry. The crowding explanation and the dry-air explanation do not compete so much as overlap, and the studies above rarely separate them.

What this does not tell you

The evidence has clear edges.

• No trial in people. The sources here are population models, guinea pigs, mice, and cells or tissue in a dish. None randomized people to a different home humidity, room temperature, or way of breathing.
• No tested interventions. Because of that, the lab findings do not tell us whether changing the humidity in your home changes your own risk. That question needs studies in people, and these papers are not that.
• Different viruses, different rules. Much of the work is on influenza. Rhinovirus, the common cold virus, was studied in the nose-temperature work. Respiratory viruses do not all respond to climate the same way.
• Population results are not personal ones. A model that reproduces the national flu curve describes a country. How often you get sick also depends on who you live with, your job, your age, and how much of what is going around you meet.
• Funding and models. One humidity study was partly funded by a humidification company, and the Shaman result comes from a model. Both are reasons to read these findings as hypotheses with support, not settled mechanisms.

What you can watch in your own data

The app logs local outdoor weather (pressure, temperature, relative humidity, wind speed, UV index) next to your own symptom entries and Apple Health data, and looks for correlations in your own history with Pearson correlation, ANOVA, and false-discovery-rate correction. It does not measure indoor air, which is the part of this story that changes most in winter. Its fixed list has nine symptoms (headache, fatigue, drowsiness, dizziness, anxiety, irritability, nausea, joint pain, insomnia), and cough, sore throat, and fever are not among them. It does not detect, diagnose, or predict infections, and it cannot tell you that you are getting a cold.

What you can do is look at the slow signals it already stores across a season: sleep, resting heart rate, HRV, and fatigue. If those drift over a few weeks that also happen to be cold and dry outside, that is a pattern you can bring to a doctor, or compare with the same weeks next year. The article on weather and sleep covers one of those signals in detail, and weather sensitivity: what the science says explains why a pattern in your own entries needs care before you call it a cause.