Remineralizing RO Water: What It Changes, and What It Costs

An RO membrane does not sort minerals into good and bad. It rejects almost everything dissolved in the water, so the stage that pulls out lead and nitrate takes the calcium and magnesium with it. Reverse osmosis remineralization is the attempt to put some of that back. It works, up to a point. The taste change is real and most people notice it inside a day. The health story is thinner than the marketing suggests, and the arithmetic below shows why.

What reverse osmosis remineralization actually adds back

A residential membrane rejects roughly 90 to 99 percent of dissolved solids. Feed water at 250 ppm TDS comes out at something like 5 to 25 ppm. Calcium, magnesium, bicarbonate, sodium, sulfate, nitrate: they drop together, because the membrane sorts by molecular size and charge, not by whether a mineral shows up on a supplement label. If you want the mechanism in detail, the stages, pressure and waste ratio of an RO system covers it properly.

A remineralizing stage is a small cartridge of dissolving media, placed after the membrane and the final carbon polish. Water passes through crushed calcium carbonate, or a blend of that with magnesium oxide, and picks up a little of it on the way to the faucet. In normal use that means 10 to 30 ppm, occasionally 40 to 50 with fresh media and a slow draw. You are not restoring the water. You are nudging it.

Before you read a spec sheet, know this. No NSF/ANSI standard certifies how much calcium a cartridge adds. Standard 42 covers aesthetic effects like taste and chlorine, 53 health effects, 58 RO systems as a unit, 401 emerging contaminants. None of them measure mineral addition. So a claim like “adds 20 ppm of essential minerals” is the manufacturer’s own bench test, run at a flow rate and feed chemistry they picked.

The health argument is weaker than sellers say

Run the numbers before you buy the story. The adult calcium target sits around 1,000 mg a day. Take tap water carrying 60 mg/L of calcium, which is not gentle water: as calcium carbonate equivalent that is close to 150 mg/L, about 9 grains per gallon at the standard 17.1 mg/L per grain, which the USGS calls hard. Two liters hands you 120 mg of calcium. Less than half a cup of milk. Magnesium is the more interesting case: two liters at 25 mg/L gives 50 mg against a daily target between 310 and 420 mg depending on age and sex.

So water was never the main source. It was a modest side contribution, and a remineralizing cartridge returns a slice of that slice. Anyone selling you a filter on the grounds that it replaces what RO removed is describing something the hardware does not do.

Here is where sources genuinely disagree, and the mechanism behind the disagreement matters more than picking a winner. A long line of observational research associates harder drinking water with lower cardiovascular mortality. Mainstream US nutrition guidance counts food as the dominant source and does not treat tap water as a delivery system for anything. Both positions can be held honestly at once. The hardness research is largely ecological: it compares regions, and regions differ in diet, income, smoking rates and a hundred other things that travel with the local water supply. The dietary accounting is cleaner, but it averages across a population, which hides the individual who eats almost no dairy and drinks very hard well water.

What I will not claim is that a 15 ppm cartridge changes a health outcome. Nobody has shown that, and I am not a physician or a dietitian. If you have a medical reason to track magnesium, potassium or sodium intake, that belongs in a conversation with your doctor rather than with a filter catalog.

Why the pH number moves so much

Carbon dioxide is a dissolved gas, and gases slip through an RO membrane almost untouched. Bicarbonate, which is what buffers ordinary tap water, does not. So the permeate arrives still carrying its CO2 with nothing left to hold the line, the CO2 forms carbonic acid, and the pH lands somewhere between 5.5 and 7.0. That is not contamination. It is what happens when you strip the buffer out of water and leave the acid behind.

Two practical consequences. First, the EPA secondary standard for pH is 6.5 to 8.5, an aesthetic guideline about taste and corrosivity rather than a health limit, and not federally enforceable. Permeate at 6.0 sits outside a recommendation, not outside a safety threshold. Second, home pH pens behave badly in this water. Low ionic strength gives the electrode almost nothing to read, so the number drifts and disagrees with itself between samples. Pour a glass, leave it on the counter an hour, and the reading climbs on its own as CO2 escapes. That is the water changing, not the meter lying. Either way, do not trust it to a tenth of a point.

Corrosivity is where low pH stops being cosmetic. Soft, acidic, low-TDS water is aggressive toward copper and solder. Under a sink it hardly matters: the permeate travels a few feet of plastic tubing to a dedicated faucet. Send it through the whole house and the calculation changes, which is why whole-house RO is a different decision from a point-of-use system, and why those installations treat post-membrane pH correction as required equipment rather than an upgrade.

Calcite and Corosex cartridges compared

Calcite is crushed calcium carbonate, basically marble or limestone screened to a working grain size. It dissolves in proportion to how acidic the water is and how long the water touches it, and it is self-limiting: as pH climbs toward neutral, the media stops going into solution. That built-in ceiling is why a calcite-only cartridge rarely takes you past about 7.5. Treat it as a feature, not a shortfall.

Corosex is magnesium oxide, and it is far more reactive. It raises pH quickly and adds magnesium rather than calcium, but on very low-alkalinity permeate it can overshoot past 9, which tastes soapy and slick. That is why it is almost always sold blended into calcite instead of on its own. Blend ratios differ between manufacturers, and none of them can know your feed chemistry.

Contact time explains most of the spread in reported results. A two-inch inline cartridge running at half a gallon a minute gives the water a few seconds against the media. A tank-style upflow contactor gives it minutes. Add in feed acidity, water temperature and how much media has already dissolved, and identical cartridges on two houses land on different numbers.

That last row carries the widest spread in the table. A mineral tank plumbed into a line costs several times what a retrofit cartridge does, and where you land depends on tank size and on whether a plumber sets it.

Calcite media dissolves away rather than plugging up. The cartridge never clogs and pressure never drops. It just does less every month until it does nothing, which makes it the easiest stage in the system to forget. Put it on the same calendar you use for the rest of the RO filter schedule, and replace it on time rather than on symptoms. Two smaller points. Flush a cartridge that sat wet and unused through a long vacation before you drink from it, and if you genuinely suspect bacteria in your water, that is a job for a certified lab or your county health department, not a filter change. And if a cartridge is sold with words like structured, energized or ionized, there is nothing in the claim to measure, and I would not pay extra for it.

Mineral drops, and when they make sense

Drops are concentrated magnesium and potassium salts, dosed by the pitcher or the gallon. The appeal is control. You pick the number instead of accepting whatever a cartridge happens to release that week, and you can dial it back if the taste turns chalky or bitter. Magnesium chloride is genuinely bitter once the dose climbs, and people notice it in plain water long before they notice it in coffee.

The problem is cost and friction. Per gallon you pay two to three times the cartridge route, and you have to remember every single time you fill something. Drops also cannot treat what you do not pour by hand, so the line feeding an ice maker or a fridge dispenser stays flat and slightly acidic however carefully you dose the pitcher. For a plumbed-in household the cartridge wins on both counts. Drops earn their place in one situation: a countertop RO unit with no plumbing to modify, where there is often nowhere to fit a post-stage at all.

What it does to espresso and tea

This is the strongest practical case for remineralization, and almost nobody selling it leads with it. Straight RO permeate makes bad espresso. With essentially no alkalinity there is nothing to buffer the acids coming out of the coffee, so shots pull sharp and sour in a way no grind adjustment fixes. The Specialty Coffee Association’s water standard has been revised over the years, and the target has stayed near 150 mg/L TDS with alkalinity around 40 mg/L as calcium carbonate. Untreated permeate is nowhere close.

There is a hardware issue too. Many espresso machines sense boiler level with a conductivity probe, and pure RO water conducts too little for the probe to see it. The machine either refuses to fill or overfills. Cafes solve this with RO plus a controlled add-back stage, which is the real argument for the whole approach: RO does not exist to make water empty, it exists to let you set the number yourself instead of taking whatever the utility sends this month.

Tea is more of a preference call. Very low mineral water makes black tea look pale and taste bright, sometimes thin. Higher alkalinity darkens the cup and flattens it out. Neither is wrong. Meanwhile the kettle stays clean on unremineralized RO, which some households would rather keep than trade for a rounder cup.

The yearly cost, laid out line by line

Buying a system with the stage already fitted adds roughly $50 to $150 to the purchase price, a smaller line than most people expect against the $180 to $600 band an under-sink RO system typically lands in, where the spread tracks stage count, tank size and certification. Retrofitting to a system you already own runs $25 to $60 for the cartridge, clips and tubing. Those numbers move with cartridge size and media blend, and with whether the housing is proprietary. That last one is the quiet expense: a captive housing can double what a refill costs five years from now.

drinking water = 1.5 gal/day
               = 548 gal/year

cartridge route
  2 cartridges x $22  = $44/year
  per gallon          = $0.08

mineral drops route
  0.5 mL/gal, $20 per 50 mL bottle
  100 gal per bottle  = $0.20/gal
  548 gal x $0.20     = $110/year

Count the second-order costs too. A post-stage slows tank refill slightly, since the water passes through one more restriction on the way out, and it eats four or five inches of cabinet depth you may not have. Installation is the easy part on a system already in place. I am not a licensed plumber, and the jobs worth paying for are the ones involving a new hole in a granite countertop or a drain saddle on a disposal, not a quick-connect fitting on existing tubing.

Who should skip it, and who should not

Most homes do not need this for health. I will put that as plainly as I can, because the whole category is sold on the opposite claim. Plenty of homes want it for taste, and that is a perfectly good reason to spend $44 a year on something you drink every day. The mistake is buying it as nutrition and then feeling cheated when nothing changes.

My own view: fit the cartridge, keep it on a schedule, and judge it entirely on whether the water tastes better to you after a week. If it does, $44 a year is cheap. If it does not, pull it out and stop paying for it. That is a decision you are allowed to make with your mouth rather than with a test strip.