Weather sensitivity: what the science actually says

Ask a room full of people whether the weather affects how they feel and many hands go up. Ask them to prove it and the room goes quiet. That gap — between a strong personal conviction and hard-to-pin-down evidence — is exactly where weather sensitivity lives. This article walks through what researchers have actually measured, where the evidence is solid, where it is messy, and how you can test the question on the one dataset that matters: yourself.

A note before we start: this is informational material, not medical advice. If your symptoms are severe or frightening, that is a conversation for a doctor, not an app.

What "weather sensitivity" means

"Weather sensitivity" is an everyday label, not a diagnosis. It describes people who notice their symptoms — headaches, joint pain, fatigue, low mood, poor sleep — tracking with the weather. In some countries the idea has a settled name (Japanese has 気象病, kishōbyō, "weather sickness"; Russian has метеозависимость), while English has no single common word and tends to talk about specific triggers like "barometric pressure headaches."

The core scientific problem is that weather changes many things at once — pressure, temperature, humidity, wind, light — and so does your life. Untangling a real effect from coincidence takes careful measurement, which is why individual studies disagree.

Barometric pressure and headache

The most studied link is between falling barometric pressure and headache. In a frequently cited clinic study, Prince and colleagues followed 77 migraine patients and found that about half were objectively sensitive to at least one weather factor — but that more patients believed weather was a trigger than the data supported (Prince et al., 2004). That last point is important and recurs throughout this field: belief runs ahead of measurement.

A small Japanese study by Okuma and colleagues added a more specific observation: migraines clustered when atmospheric pressure dropped roughly 6–10 hPa below standard, into the 1003–1007 hPa range (Okuma et al., 2015). A 2019 narrative review pulled these threads together and concluded that the evidence is inconsistent and fragmented — studies disagree even on the direction of the association, and there are no randomized trials of treatment for pressure-triggered headache (Current Pain and Headache Reports, 2019).

The honest summary: pressure change is a real candidate trigger for some people with migraine, not a universal switch.

Joints: strong belief, weak evidence

Joint pain is where personal conviction is strongest and the aggregate evidence is weakest. The most rigorous look so far is a 2024 systematic review with meta-analysis of case-crossover studies covering more than 15,000 people — a design that compares each person to themselves on pain days versus pain-free days, which controls for a lot of noise. Across rheumatoid arthritis, knee pain and low back pain, it found no association between pain and humidity, air pressure, temperature or precipitation. The one clear exception was gout, where hot, dry spells were linked to flares (Seminars in Arthritis and Rheumatism, 2024). The takeaway is not that weather never matters to anyone, but that for the common joint complaints the average effect is close to undetectable — a striking gap from how strongly people feel it.

Sleep, heat and energy

Weather also acts on you indirectly, through sleep. Using data from hundreds of thousands of U.S. residents, Obradovich and colleagues showed that warmer-than-normal nights measurably increase reports of insufficient sleep, with the effect concentrated in summer and among older and lower-income people (Obradovich et al., 2017). Poor sleep, in turn, is one of the most reliable amplifiers of pain, low mood and low energy. So a hot, humid spell can leave you feeling wrecked without any exotic mechanism — it simply stole an hour of your sleep.

Why your memory is a bad judge

If the effects are often small, why is the belief so strong? Two well-known quirks of human cognition. Confirmation bias: a headache on a stormy day is filed as evidence, a headache on a clear day is forgotten, and a clear day with no headache is never counted at all. Recall bias: we reconstruct the past to fit the story we already hold. This is not a character flaw; it is how memory works. It is also precisely why the Prince study found more believers than the data supported.

The way out is not to argue with your memory but to replace it with a record.

Finding your own pattern

Population studies answer "does weather affect people on average?" They cannot answer "does weather affect me?" For that you need your own data, collected the same way a study would: consistently, over enough time, with the weather logged alongside the symptom.

Three principles make a personal record trustworthy:

1. Log misses, not just hits. A note that says "clear day, felt fine" is as valuable as "storm, bad headache." Without the misses you are just repeating confirmation bias with extra steps.
2. Give it time. A handful of days proves nothing. Weeks to months let genuine patterns separate from coincidence.
3. Let statistics judge, not vibes. Correlation methods can flag when an association is stronger than chance would produce — and, just as usefully, when it is not.

This is the reasoning behind how MeteoHealth is built. It records your symptoms and mood in two taps, pairs them with local weather and the Apple Health metrics you already collect, and runs statistics — Pearson correlations, ANOVA, false-discovery-rate correction — on your device to show which associations in your data look real. It does not diagnose or predict; it shows you your own pattern, with honest uncertainty, and leaves the conclusions to you and your doctor.

The bottom line

Weather can plausibly nudge headaches through pressure changes; its average effect on the common forms of joint pain is close to nil, gout aside; and it can clearly disrupt sleep and therefore energy. The effects that are real tend to be small, highly individual, and easy to overestimate from memory alone. The scientifically honest move — and the practical one — is to stop guessing and start measuring your own signal.