The gap between UF and RO
Australian households are usually offered two membrane options: ultrafiltration, which is a physical barrier to organisms and particles but does nothing to anything dissolved; and reverse osmosis, which strips out very nearly everything including the minerals you might have wanted to keep.
Nanofiltration sits deliberately between them. NF membranes have a molecular weight cut-off of roughly 200 to 1,000 daltons, which makes them very good at rejecting larger organic molecules and multivalent ions — calcium, magnesium, sulfate — while allowing a significant share of monovalent ions such as sodium, potassium and chloride to pass.
That selectivity is not an accident of manufacturing tolerance; it is the design goal. Doubly charged ions feel a stronger electrostatic interaction with the membrane and are rejected hard. Singly charged ions feel a weaker one and find their way through.
What that selectivity buys
| Target | Nanofiltration | Reverse osmosis |
|---|---|---|
| Hardness (calcium, magnesium) | High rejection | High rejection |
| Sulfate | High rejection | High rejection |
| Sodium, chloride | Partial — much passes | Near-complete rejection |
| Long-chain PFAS (PFOA, PFOS) | 90–99% | 90–99%+ |
| Short-chain PFAS (PFBA, PFBS) | Weak point — can slip through | Better |
| Larger organic molecules | High rejection | High rejection |
| Operating pressure | Lower | Higher |
| Water recovery | Higher — less to drain | Lower |
| Seawater desalination | Not suitable alone | The standard |
Peer-reviewed comparison work supports the picture: in water reuse applications, rejection of inorganics by a tight NF membrane was similar to RO except for monovalent ions, and rejection of trace organic contaminants was similar between the two. The published energy figures also favour NF, with specific energy consumption for brackish groundwater reported in the range of roughly 0.2 to 3.5 kWh per cubic metre depending on feed and target concentration.
One nuance worth knowing: “nanofiltration” is a band, not a single specification. Tighter NF membranes behave almost like RO for monovalent ions; looser ones let most monovalent salts through and concentrate on hardness and organics. Two products both labelled NF can behave quite differently, which makes the specific membrane grade the thing to ask about.
Softening without salt — the interesting case
The obvious Australian application is hardness. NF rejects calcium and magnesium at high rates while leaving much of the sodium and chloride alone — so it softens water without the sodium trade that ion exchange requires.
Set against the alternatives from our other reviews: ion exchange removes hardness but adds roughly 0.46 mg/L sodium per mg/L of hardness removed, consumes salt and discharges brine. Salt-free conditioning leaves the hardness in the water entirely and only changes how it behaves. NF genuinely removes the hardness minerals, adds no sodium, and uses less pressure and less wastewater than RO.
On paper that is an attractive middle ground, and for a household in a very hard Perth zone it sounds close to ideal. Which raises the obvious question.
So why isn’t it in Australian homes?
Because the economics do not scale down. The assessment in the technical literature is direct: the capital cost of membranes, pressure vessels and monitoring instrumentation is hard to justify at household scale, and proven point-of-use options — RO and activated carbon — deliver strong protection for a fraction of the installed cost at a single tap or for whole-home volume.
NF is genuinely mainstream, but at municipal and commercial scale: softening plants, water reuse, groundwater treatment, and as pre-treatment ahead of RO to strip hardness and organics before they foul the expensive membrane.
This creates a specific claim to check. Because “nanofiltration” sounds more advanced than RO and carries a mineral-retention story that markets well, the word appears on domestic products whose actual membrane specification is unclear. If a system is sold to you as nanofiltration, ask for the membrane model and its molecular weight cut-off, and what it rejects at your feed pressure. “Nano” in a product name is not a specification.
What nanofiltration cannot do
It is not a disinfection barrier. NF membranes remove most bacteria and some viruses, but they are not intended as standalone microbiological treatment and carry no such certification. On untreated supply they are paired with UF, UV or chlorination.
It will not desalinate seawater. Monovalent rejection is too low; seawater needs RO. NF is sometimes used as pre-treatment ahead of it.
Short-chain PFAS are a real weakness. PFBA and PFBS have smaller molecular dimensions and rejection drops. If short-chain PFAS is your actual concern, RO or a dedicated adsorption stage is the answer — see our PFAS guide.
Chlorine destroys the membrane. Polyamide thin-film composite membranes typically tolerate free chlorine below about 0.1 ppm — far below what Australian mains supply carries. Carbon pre-treatment is mandatory, not optional, exactly as with RO.
Fouling and scaling are ongoing. Field reviews of membrane plants report declining water recovery, fouling, scaling and concentrate disposal as the recurring practical problems, alongside a need for technical knowledge that household owners generally do not have.
Our verdict on the evidence
Nanofiltration is a well-characterised, genuinely useful technology that most Australian households will never appropriately buy. The selectivity is real and the peer-reviewed comparisons are favourable: near-RO rejection of hardness, sulfate, organics and long-chain PFAS, at lower pressure, with better recovery and less water to drain, while leaving sodium and potassium largely in place.
Its problem is not performance but fit. At municipal and commercial scale it is mainstream and sensible. At household scale the instrumentation and capital cost are hard to justify against an RO system that costs a fraction and does more — and the mineral-retention advantage, while genuine, is a taste and preference argument rather than a health one.
Our position: if you are offered domestic nanofiltration, treat the word as a prompt for questions rather than a specification. Get the membrane grade and cut-off, confirm carbon pre-treatment is fitted, ask what it does about short-chain PFAS if that matters to you, and compare the installed and running cost honestly against RO. If the answer to any of those is vague, you are being sold a name.
Primary sources: peer-reviewed comparative analysis of reverse osmosis and nanofiltration for removal of dissolved contaminants in water reuse applications (Desalination), finding similar inorganic and trace organic rejection between NF90 and RO except for monovalent ions; comparative study of community RO and NF systems for total hardness removal in groundwater, including reported recovery decline, fouling, scaling and concentrate disposal problems; published partial-desalination comparison of NF, brackish and seawater RO and membrane capacitive deionisation reporting NF specific energy consumption of approximately 0.2–3.5 kWh/m³; published NF membrane characterisation giving molecular weight cut-off of roughly 200–1,000 daltons and divalent versus monovalent rejection behaviour; polyamide thin-film composite free chlorine tolerance below 0.1 ppm. Reviewed August 2026. FilterOut takes no commissions and carries no paid placements — see who to trust for filter advice.
FAQ
What is the difference between nanofiltration and reverse osmosis?
Both are pressure-driven membranes, but NF is looser. It strongly rejects multivalent ions — calcium, magnesium, sulfate — and larger organics, while letting much of the sodium, potassium and chloride pass. RO rejects nearly everything dissolved, at higher pressure and with more water sent to drain.
Does nanofiltration soften water?
Yes, selectively. It rejects the calcium and magnesium that cause hardness at high rates while leaving more of the monovalent minerals in place — so it softens without the sodium that ion exchange adds. That selectivity is the main reason NF exists.
Does nanofiltration remove PFAS?
Long-chain PFAS such as PFOA and PFOS, commonly at 90–99%. Short-chain compounds like PFBA and PFBS are a known weak point because their smaller molecular size lets them slip through. If short-chain PFAS is the concern, reverse osmosis or a dedicated adsorption stage is the better answer.