People in Britain's hard-water areas had slightly smaller brains on MRI. The researchers expected the opposite, and the link weakened when they allowed for where people lived

Hard water is water carrying a lot of dissolved calcium and magnesium, picked up from chalk and limestone on its way to the tap. It is what leaves the white crust in kettles and on shower heads. For decades it has had a reasonably good name in health research: studies of whole populations have tended to find slightly less heart disease and high blood pressure where the water is hard. So when a radiology group in Beijing went looking at hard water and the brain, they expected good news. Their stated hypothesis was that harder water would go with larger, better-preserved brains.

They used the UK Biobank, which recruited half a million middle-aged Britons between 2006 and 2010 and later scanned the brains of tens of thousands of them. Each person's water hardness was taken from the 2005 figures for the water supply zone of their home postcode, as reported by the water companies. The brain measure was volume on a single MRI scan: whole brain, white matter, grey matter and 122 smaller regions, each scaled for head size. After people with missing data or a neurological diagnosis such as stroke, dementia or multiple sclerosis were left out, 34,183 remained. The paper was published in Health Data Science on 6 October 2026.

The result went the other way. Compared with people on soft water (up to 60 mg of calcium carbonate per litre), people on the hardest water (over 180 mg/l) had slightly lower total brain volume, white matter and grey matter. Calcium on its own showed the same pattern. Magnesium did not: more of it went with a little more grey matter, up to about 4.5 mg/l, after which the curve turned down again. The links were stronger in people over 60.

How big is 'slightly'? The difference in total brain volume for the hardest-water group was 0.06 of a standard deviation. The authors translate that as roughly 0.6% of brain volume and compare it to one to two years of normal brain ageing. That comparison is theirs, and it is a fair way to make a small number feel real, but it is still a small number, measured once, in a population where brain size varies a great deal from person to person.

The pattern across the four hardness groups is also less tidy than the headline suggests. The moderately hard group (60 to 120 mg/l) had slightly smaller brains than the soft-water group. The hard group (120 to 180 mg/l) had slightly larger ones, though that difference could be chance. Then the very hard group was smaller again. If calcium in water were steadily shrinking brains, you would expect each step up to look worse than the last.

The bigger issue is geography. Water hardness in Britain is set by the rocks under each region, so it maps closely onto where people live, and where people live comes with its own air, jobs, incomes, accents, hospitals and scanners. The table of participants shows it: the hardest-water group had more people in the lowest deprivation band (27% against 18%), more people with high incomes (12% against 7%) and more than twice the share of non-White participants (5.7% against 2.0%) compared with the soft-water group. The study adjusted for income, deprivation, education, ethnicity, smoking, drinking, exercise, weight, diet and more. But the authors also report that when they adjusted for which UK Biobank assessment centre each person attended, a rough stand-in for region, the associations weakened. They chose to leave that adjustment out, arguing it overcorrects because hardness and location are so tightly linked, and they accept that 'residual spatial confounding cannot be ruled out'.

That choice is defensible, but it cuts both ways. If an adjustment for region shrinks the link, the study cannot easily tell apart 'hard water does something to the brain' and 'people in hard-water regions differ in ways the other adjustments did not capture'. The paper also had no data on other things in the water, such as fluoride, sulphate or industrial contaminants, nor on whether people drank tap water, filtered it or used a softener at home. The water figure is for one year, 2005, applied to a scan taken about a decade later.

Our reading: this is a large, carefully described data analysis that turned up an unexpected and small association, and the authors are right to call for it to be repeated. It does not show that hard water harms the brain, and the authors' own suggestion that drinking-water minerals could be adjusted 'for neuroprotection' runs well ahead of what one observational study can carry. If you live in a hard-water area, nothing here is a reason to switch to bottled water or install a softener for your health. Calcium and magnesium from water are a small part of what most people get from food, and the earlier heart research, also observational, pointed in the opposite direction.