What Is Left in Water After Reverse Osmosis?

After reverse osmosis, the water you drink is the permeate (product water): low in total dissolved solids (TDS) compared with the feed, but not empty of everything. Typical RO membranes reject a large share of dissolved salts — often in the mid-to-high 90s percent range under manufacturer test conditions — but rejection is not 100%. What still shows up depends on the membrane, feed chemistry, pressure, temperature, and whether the unit includes carbon or other stages. Dissolved gases such as carbon dioxide commonly pass; some small molecules and a residual fraction of salts can pass too. The stream that goes to the drain is concentrate (reject/brine), not what you drink.

This guide answers the drinking-water question first, then covers remineralization and “alkaline RO” marketing, and only after that the wastewater side. I am a homeowner researching agency and manufacturer documentation — not a water-treatment contractor — and I have not sampled your tap. Performance numbers below come from cited EPA, CDC, NSF, or manufacturer sources. This guide was substantially rewritten and re-fact-checked in August 2026.

Key takeaways

• RO drinking water is low-TDS permeate. It is not “pure H2O,” and RO does not remove “everything.”
• Under EPA WaterSense’s summary of NSF/ANSI 58, certified systems must show at least 75% TDS reduction as a baseline — real membranes often do better, but there is no universal percent for every contaminant.
• Dissolved gases (notably CO2; also hydrogen sulfide when present) commonly pass RO membranes and can affect taste, odor, pH, and conductivity.
• Remineralization and “alkaline” add-ons change taste and mineral profile; they are not a magic health upgrade. Per WHO, food — not drinking water — is the principal mineral source for most people.
• Concentrate/wastewater is real and often substantial on under-sink systems — a separate issue from what remains in the glass.

What “left in the water” actually means

An RO system splits feed water into two streams. The EPA’s overview of drinking-water treatment technologies describes it this way: pressure forces water through a semi-permeable membrane; the portion that passes is permeate, along with lower-molecular-weight substances the membrane does not fully stop; the portion that does not pass is concentrate, which retains higher-molecular-weight substances and many undesirable contaminants.

• In the glass (permeate): mostly water, plus whatever the membrane and any pre/post filters did not remove under your operating conditions.
• In the drain (concentrate): rejected salts and contaminants, concentrated into less water.

This page focuses on the glass. The drain stream matters for water bills — covered briefly below — but it is not what most people mean by this question.

RO permeate is low-TDS water — not mineral-free magic

TDS is a catch-all for dissolved salts, minerals, metals, and other dissolved matter. The CDC’s home water-treatment overview says reverse osmosis removes parasites, bacteria, and viruses, and removes some chemicals including lead, copper, chromium, chloride, and sodium. It may also reduce arsenic, fluoride, radium, sulfate, calcium, magnesium, potassium, nitrate, and phosphorus. The CDC’s practical rule is the right one: check the system’s label for the specific chemicals it is certified to reduce.

Contaminant reduction claims for point-of-use RO systems live under NSF/ANSI 58 (materials safety, structural integrity, TDS reduction, efficiency/recovery ratings, and elective claims such as lead, fluoride, nitrate/nitrite, VOC reduction, and others). A claim that is not on the certified label is advertising, not a verified performance statement.

DuPont’s FilmTec RO/NF technical manual describes the membrane as generally acting as a barrier to dissolved salts and inorganic molecules, and to organic molecules with a molecular weight greater than about 100, while water passes. It states that rejection of dissolved salts is typically 95% to greater than 99%, depending on membrane type, feed composition, temperature, and system design. That is a manufacturer range under defined conditions — not a promise that your under-sink unit hits 99% on every ion in your water.

Because membranes are imperfect barriers, some dissolved solids always remain in permeate. “Zero TDS” marketing is usually a meter quirk, a polishing stage, or exaggeration. Absolute permeate TDS also scales with feed TDS: the same percent rejection leaves more behind when the incoming water is saltier.

What may still be in RO drinking water

A practical inventory for the glass — possibilities, not a scare list every home will see.

1. Residual dissolved salts and minerals

RO reduces calcium, magnesium, sodium, and many other ions; it does not delete them with certainty. EPA WaterSense’s performance overview states that NSF/ANSI 58–certified systems must reduce TDS by at least 75% as a baseline. Many residential membranes advertise higher salt rejection under their own test conditions, but that floor already expects leftover dissolved solids.

How much remains depends on feed quality, membrane age/fouling, pressure, temperature, and clogged prefilters. A tired membrane can pass more TDS than the brochure suggests.

2. Dissolved gases

This is the category that trips people up. The FilmTec technical manual is explicit: carbon dioxide is not rejected by the membrane and appears in product water, where it forms carbonic acid and can raise conductivity. The same manual notes that gases including hydrogen sulfide (H2S) typically pass RO and NF membranes — which is why industrial systems often strip H2S from permeate when it is present.

• Flat or slightly acidic-tasting water (CO2 / lower pH) — the EPA’s treatment-technologies overview also notes that RO can lower treated-water pH and may need post-treatment for corrosion control in larger systems.
• Rotten-egg odor if sulfide is in the source — an RO membrane alone is the wrong tool; you need sulfide-specific treatment, not a bigger “rejection %” sticker.

3. Some small organic molecules — claim- and condition-dependent

Some articles claim RO always “leaves behind” herbicides, pesticides, and VOCs. That is too blunt. EPA WaterSense lists VOCs among contaminant types RO systems can potentially address, and NSF/ANSI 58 includes elective VOC reduction claims for systems tested for them. DuPont’s guidance that organics above roughly MW 100 are generally rejected also implies smaller, more mobile organics are likelier to challenge a membrane-only barrier.

• Do not assume a bare membrane rejects every pesticide or VOC in your water.
• Do not assume the opposite — many under-sink RO units include carbon stages, and some carry certified VOC claims.
• Match certified claims on the box to your water report. No listing → treat the claim as unproven for that unit.

If your main concern is chlorine taste on municipal water and you do not need broad dissolved-solids reduction, a certified pitcher or faucet filter may be enough — see our water filter pitchers guide. Pitchers are not RO, and they are not a substitute when you need dissolved-metal or certified RO claims.

4. Storage tank and plumbing effects

Permeate sitting in a storage tank can pick up taste from tank materials, stagnant water, or a neglected post-carbon stage. That is maintenance, not “the membrane failed.” Follow the filter-change schedule, and flush a few glasses after long unused stretches.

High rejection is not 100% rejection

People hear “99% rejection” and hear “everything gone.” That is not what manufacturers or standards bodies mean.

Number you see	What it usually means	What it does not mean
NSF/ANSI 58 ≥75% TDS reduction (baseline)	Minimum ionic-rejection integrity for certified POU RO, per EPA WaterSense citing NSF/ANSI 58	That every contaminant is cut by 75%, or that 25% of “toxins” remain as a health metric
Manufacturer “typically 95% to >99%” salt rejection	Dissolved-salt performance under stated membrane test conditions (e.g., FilmTec manual)	A universal percent for lead, nitrate, fluoride, PFAS, or VOCs on your tap
Elective NSF/ANSI 58 contaminant claims	That specific contaminant was tested to the standard’s protocol	That unlisted chemicals are covered

I am deliberately not reprinting a homemade percent-rejection table for dozens of chemicals. Those tables circulate online with no method notes. For a specific solute, use the membrane maker’s data under stated conditions or the unit’s NSF/ANSI 58 listing — not a blog collage.

Remineralization and “alkaline RO” marketing

RO permeate often tastes “flat” because hardness minerals are reduced and CO2 can lower pH. Remineralization cartridges (calcium/magnesium salts, calcite, and similar) and alkaline add-ons put minerals back and raise pH for taste. That can be a fair preference, and post-treatment can make very soft permeate less aggressive to metal plumbing in some setups.

• Not a cure for missing certified claims. Adding minerals does not fix arsenic, lead, or nitrate if the RO stage is not reducing them.
• Not proof RO water is “unhealthy” without them. WHO’s review of calcium and magnesium in drinking-water states that food is the principal source of both minerals; for most people on a normal diet, drinking water is a minor contributor by comparison. Treating remineralization as a medical necessity is marketing.
• Not “alkaline water heals X.” Raising pH changes chemistry in the glass; it does not turn a filter into a therapy product.

My honest read: buy remineralization if you dislike low-mineral taste or your installer recommends it for plumbing. Skip it if taste is fine and certified contaminant reduction is what you actually need.

The wastewater side (after the drinking-water answer)

Concentrate is the other half of RO. On point-of-use systems, EPA WaterSense reports that a typical unit sends about five gallons or more of reject water down the drain for every gallon treated — inefficient units can reach about ten to one. WaterSense-labeled models must demonstrate no more than about 2.3 gallons of drain water per gallon treated, while still meeting NSF/ANSI 58–tied performance criteria (including the 75% TDS baseline and verified elective claims).

That reject stream is where most removed dissolved solids go — the cost of producing low-TDS permeate, not “toxins left in your drinking water.” If waste is your main objection, compare efficiency ratings, prefer WaterSense-labeled POU RO when available, or ask whether a non-RO filter certified for your specific problem would do with less waste.

How to check your own RO water

1. Start with a water report or certified lab test for the contaminants you care about — especially on a private well.
2. Match NSF/ANSI 58 claims (and any carbon-stage claims) to that list. No claim on the label → no verified promise.
3. Spot-check TDS on feed vs permeate for a rough membrane health check. A sudden rise in permeate TDS is a maintenance clue, not a full chemical diagnosis.
4. Treat odor separately. Sulfide smell after RO points to gas passage or source chemistry, not “buy a higher GPD membrane.”

When to call a pro

Call a licensed plumber or reputable water-treatment professional for under-sink RO on a shared drain, low feed pressure needing a correctly sized booster, well water that needs whole-house pretreatment before POU RO, or lab results showing a health-based contaminant you are not equipped to interpret. I research manuals and agency pages; I do not design treatment trains for contaminated wells from a blog comment.

How this was researched

Claims above were checked in August 2026 against the EPA drinking-water treatment technology overview, EPA WaterSense POU RO materials (including the performance overview documenting the NSF/ANSI 58 75% TDS baseline), the CDC home treatment systems page, NSF’s NSF/ANSI 58 public overview, DuPont’s FilmTec RO/NF technical manual for salt rejection and dissolved-gas behavior, and WHO’s calcium/magnesium-in-drinking-water review for the food-as-principal-mineral-source framing. I did not lab-test residential RO brands for this rewrite, and I did not invent per-contaminant rejection percentages.