Is reverse-osmosis or distilled water too pure to drink?

What RO takes out. An RO system pushes water through a membrane that holds back dissolved solids. The CDC lists RO as removing parasites, bacteria and viruses and some chemicals, including lead, copper and chromium, and says it may reduce arsenic, fluoride and nitrate³. A review of point-of-use treatment found that carbon, ion exchange and RO are all used against PFAS (long-lasting industrial “forever chemicals”), but real-world performance, especially for short-chain PFAS, is poorly characterized⁴. Look for NSF/ANSI 58, the RO standard, and for the specific contaminants listed on the label⁵; see choosing a filter.

Where minerals come from. A 1981 study noted that only 10–20% of daily calcium and magnesium intake comes from drinking water¹. The BfR estimated that bottled mineral water supplies about 30 mg of magnesium a day to the median German consumer, 7–8% of total intake². Tap water varies: in half of 21 North American supplies, 2 liters a day gave 8–16% of the calcium and 6–31% of the magnesium reference intake⁶. RO removes that contribution, and food has to cover it. Cooking in soft water also lowered the calcium in potatoes and vegetables, while hard water raised it¹.

What the agencies concluded. WHO's 2005 monograph, Nutrients in drinking-water, examined the long-term effects of demineralized water⁷. In 2020 Germany's Federal Institute for Risk Assessment (BfR) re-reviewed the evidence. It called the 2005 chapter's statements outdated, since they rested on older, mostly Russian-language studies, and noted that later WHO guidance found too little data to propose minimum or maximum mineral levels. Its conclusion: with a balanced diet and normal drinking amounts, no adverse effects are discernible. It advised extra attention to calcium and magnesium for people with very unbalanced diets, vegans and others avoiding dairy, and pregnant or breastfeeding women who rely mainly on very low-mineral water².

Observational signals. Pooled studies link higher magnesium in drinking water to lower cardiovascular mortality⁸, but a large Dutch cohort found no overall link⁹, and regional differences in diet and lifestyle could explain the pooled finding.

The waste trade-off. A typical point-of-use RO unit sends five gallons or more of reject water down the drain per gallon treated; WaterSense-labeled models must send 2.3 gallons or less¹⁰.

Pooled case-control data. Catling and colleagues' systematic review included 14 studies (9 case-control, 5 cohort). Pooling seven case-control studies, higher magnesium in drinking water went with lower cardiovascular mortality (odds ratio 0.75, 95% CI 0.68–0.82). Evidence for calcium was unclear, and the cohort studies found no association between water hardness and cardiovascular or stroke mortality⁸. People in different areas differ in diet, income and smoking, and exposure is typically assigned by place of residence.

A large null cohort. Leurs and colleagues linked tap-water hardness, calcium and magnesium in every Dutch postal code to mortality in a cohort of 120,852 people aged 55–69 followed for 10 years (a case-cohort analysis of 1,944 ischemic heart disease deaths, 779 stroke deaths and 4,114 subcohort members). Hard versus soft water showed no relation to ischemic heart disease mortality (hazard ratio 1.03, 95% CI 0.85–1.28 in men; 0.93, 0.71–1.21 in women) or stroke. Only in men in the lowest fifth of dietary magnesium intake did higher tap-water magnesium go with less stroke death (HR 0.75 per mg/L, 95% CI 0.61–0.91), and women showed the opposite pattern⁹.

A natural experiment. After Israel began supplying desalinated seawater, which is nearly free of magnesium, Koren and colleagues examined serum magnesium in 66,764 people. In communities that switched in 2013, mean serum magnesium slipped from 2.065 to 2.057 mg/dL, and the share at or below 1.6 mg/dL rose from 1.62% to 2.01%, a 24% relative increase; communities that did not switch showed no such change¹¹. The average shift is tiny, and whether the rise in low values matters clinically is unknown.

Children. Huang and colleagues compared 10–13-year-olds who had drunk very low-mineral water (conductivity 40 µS/cm; 431 children) or normal-mineral water (345 µS/cm; 229 children) at school for four years, in a retrospective cohort. The very-low-mineral group had a smaller height gain (16.6 versus 22.3 cm), lower bone mineral content (1.92 versus 2.14 kg) and a higher bone-resorption marker¹². The BfR noted that total daily calcium and magnesium intakes in both groups were well below recommendations (about 36–41% of the calcium and 82–86% of the magnesium recommended), so the study may reflect poor diets as much as water².

The agency review. The BfR weighed animal studies (rating a long-term rabbit study as low in scientific relevance for design reasons), human intervention studies (judged inadequate as a basis for risk assessment) and epidemiology. It concluded that the WHO position stands: the evidence on magnesium or hardness and cardiovascular mortality is debated and does not prove causality².

Null and conflicting results. The pooled association⁸ and the small magnesium shifts after desalination¹¹ point one way, while the large Dutch cohort⁹ and the BfR review² point the other. A causal reading remains unproven, and we found no trial testing RO water against mineral water on health outcomes.

Bottom line (confidence 3 of 5). For healthy adults eating a varied diet, RO or distilled water is probably fine to drink; the minerals lost are small compared with what food provides. The people to think about are those with poor or restricted diets, pregnant or breastfeeding women, children with low mineral intakes, and endurance athletes drinking only very low-mineral water. Remineralizing is a reasonable preference, not a proven necessity.