Radium and uranium in city water — four standards, one certified filter claim

By Andrew Bates, FounderPublished 2026-09-11Updated 2026-09-11

EPA regulates four radionuclides — gross alpha at 15 pCi/L, combined radium at 5 pCi/L, beta and photon emitters at 4 mrem/yr, and uranium at 30 µg/L. Only radium has a certified household claim. Uranium, gross alpha and beta emitters have none under any NSF standard.

Four standards, and what each one is

Radionuclides are the group most people skip on a water report, partly because the units are unfamiliar.

Regulated radionuclide	EPA limit	Health goal
Gross alpha particle activity	15 pCi/L	zero
Combined radium-226 and radium-228	5 pCi/L	zero
Beta particle and photon emitters	4 mrem/yr	zero
Uranium	30 µg/L	zero

Every goal is zero. That is a health goal, not the standard. The enforceable figure is the limit in the middle column.

EPA names the source of radium and uranium as erosion of natural deposits. These are geology, not industry.

The gap that matters when you go shopping

One of those four has a certified household product. Three do not.

NSF offers a radium 226/228 reduction claim under exactly two standards. There is no selectable claim for uranium, gross alpha, or beta and photon emitters anywhere in its listing.

That was checked against the full claim list, with radium as the control: radium returns two options, so the search works, and the uranium absence is real rather than a failed query.

→ The regulated contaminants with no certified filter at all

An absent claim is not proof that no technology works. EPA and state health departments both name treatments for uranium. It means nothing has been tested and listed, so nothing can be verified before you buy.

Radium has two routes, and they are opposites

This is the useful part, and it is not obvious from any product page.

NSF/ANSI 44 — cation exchange softeners. 114 certified models from 8 companies. Every one is listed point-of-entry, treating the whole house.

NSF/ANSI 58 — reverse osmosis. 120 certified models. Every one is point-of-use, treating one tap.

Same contaminant. Two certifications. Opposite architectures, and opposite costs.

→ Reverse osmosis: what NSF/ANSI 58 actually certifies

A softener that softens is not a softener certified for radium

Under NSF/ANSI 44, the radium and barium claim sets are the same 114 models. Both sit inside the 192 models certified for hardness reduction.

Which leaves 78 hardness-certified softeners holding neither claim.

So “it’s a softener, softeners handle radium” does not survive contact with the listing. Ion exchange is the mechanism either way. Only 114 models have had that performance tested and published.

→ Water hardness and softener sizing

Radon is not on the list at all

Worth saying plainly, because the word gets used loosely.

EPA states the gross alpha standard covers all alphas not including radon and uranium. A radon-in-water rule was proposed and remains proposed.

So your utility has no enforceable radon figure to meet, and none to report. Radon in indoor air is a separate matter with its own testing route, and it is generally the larger exposure.

Why your report’s number may be years old

Radionuclide monitoring relaxes faster than most people expect.

After four consecutive quarters of initial sampling, EPA sets the repeat frequency by result:

Initial result	Then sampled
Below the detection limit	Once every 9 years
At or below half the limit	Once every 6 years
Above half the limit, at or below it	Once every 3 years
Above the limit	Back to quarterly until four are clear

All samples are taken at each entry point to the distribution system — not at anybody’s tap.

A clean radionuclide line is meaningful. It is also, on the relaxed schedule, potentially a measurement from several years ago.

→ How to read your water quality report

How common is this actually

Rare, and geographically clustered.

Minnesota Department of Health reports that in 2014, nine of Minnesota’s municipal systems — 1.2 percent — had radium above the EPA standard. The highest source levels sit in the Mount Simon-Hinckley and Jordan aquifers in the southeast.

MDH also notes surface water generally does not contain radionuclides at levels of concern. This is a groundwater question.

Nationally, EPA estimated when it made the rule that only 795 systems would have to install treatment.

The decision framework

1. Read the four lines before anything else

They are on your Consumer Confidence Report, because community systems must test for radionuclides and publish what they find.

2. If every figure is under the limit, stop

There is no product decision here. The compliance question has been answered by someone else, at their expense.

3. If radium is over, choose the architecture first

Whole-house softening and point-of-use reverse osmosis are both certified routes. Which one fits depends on whether you want treated water everywhere or at one tap.

→ Point of use or whole house

4. If you pick a softener, search the radium claim by name

Only 114 of 192 hardness-certified models hold it. The word “softener” on a box does not carry it.

→ How to check a certification claim yourself

5. If uranium is your number, expect no certified product

Nothing holds a listed uranium claim. Take that to your utility and your state drinking water programme rather than to a retailer.

The decision path

Your situation	What decides it	Where that lands you
Have not read the report	Four lines, published annually	Read it first
All four under the limit	Nothing to treat	No purchase
Radium over, want whole-house	114 certified point-of-entry softeners	NSF/ANSI 44, claim-checked
Radium over, want one tap	120 certified RO systems	NSF/ANSI 58, claim-checked
Already own a softener	78 of 192 hold no radium claim	Check the listing for your model
Uranium over the limit	No certified claim exists	Utility and state, not retail
Gross alpha or beta over	No certified claim exists	Utility and state, not retail
Worried about radon	Not regulated in water	Test indoor air instead
On surface water	Generally not at levels of concern	Very low priority

Choose to act if

A radionuclide line on your report is at or above the limit.
Your system draws from groundwater in an area known for radium, and the figure sits near the limit on a multi-year sampling schedule.

Choose to wait if

The figures are well under the limits and the system is on surface water.
You already own a softener and have not yet checked whether your model holds the radium claim. That check is free.

Choose to do nothing if

Every figure is under the limit and nothing about your geology or your report suggests otherwise. Compliance is the utility’s obligation, not yours.

Mistakes that cost the most money

Buying for radon. It is not regulated in water and your report will not carry it. The exposure that matters is usually indoor air.

Assuming a softener covers radium. 78 of 192 hardness-certified models hold no radium claim. Check the model, not the category.

Buying a uranium filter. No certified uranium claim exists under any NSF standard, so nothing on the market has listed uranium performance.

Reading MCLG zero as a failure. Every radionuclide has a goal of zero. It is a health goal, and no utility is judged against it.

Treating one clean line as current. On the relaxed schedule, that sample may be up to nine years old.

Skipping the report and buying on geography. Exceedance is uncommon even in affected states — 1.2 percent of Minnesota’s systems in 2014.

Three situations, and what changes

A groundwater system in southeastern Minnesota, radium at 4 pCi/L. Under the limit, so the utility is compliant and no treatment is required of it. On a three-year sampling cycle, a figure this close to the limit is worth watching in next year’s report. If you act, both certified routes are open.

A large surface-water system, all four lines low. MDH states surface water generally does not contain radionuclides at levels of concern, and the report confirms it. Nothing here justifies a purchase, and a salesman raising radiation has no number to point at.

Uranium above 30 µg/L. The one case where the market has nothing certified to offer. EPA names kidney toxicity alongside cancer risk for long-term exposure above the limit. This is a conversation with the utility and the state drinking water programme, and the utility is already required to act.

The objections worth answering

“Any radiation in water is unacceptable.” EPA sets every radionuclide goal at zero for exactly that reasoning. It then sets enforceable limits at what is feasible. MDH notes about half of annual radiation exposure comes from natural sources such as sun, soil and rock.

“Surely a good filter handles all of it.” Only radium has a certified claim. For uranium, gross alpha and beta emitters there is nothing listed to compare, so no product can be verified against them.

“My softener already deals with it.” Only if your model is one of the 114. Ion exchange is the right mechanism, but certification is per model and 78 hardness-certified softeners do not carry it.

“The utility would have told me.” It does, annually, in the report. That is what the radionuclide lines are. The reporting duty is met by publication, not by a phone call.

“Reverse osmosis is the safer choice for this.” For radium it is one of two certified routes, and it treats one tap rather than the house. A point-of-entry softener and a point-of-use RO system solve different halves of the problem.

Who this is for

Households on a public water system who have found radium, uranium, gross alpha or beta emitters on their annual report and want to know what, if anything, follows.

It is also for anyone being sold a device on the word “radiation” who would like to see whether a certified claim exists for the contaminant in question.

Private wells are the harder case, because nobody monitors them.

→ Well water: the six-parameter test panel

What to settle first

Which of the four lines is actually over, and does a certified claim exist for it?

The first is free in your report. The second is free in NSF’s listing. Only one of the four has an answer that ends in a product.

→ How to read your water quality report → The regulated contaminants with no certified filter at all → Which claims exist at all, and how populated each is → Reverse osmosis: what NSF/ANSI 58 actually certifies → City water treatment, and the three gaps a utility cannot close → Radon in water, the radionuclide with no limit at all → Distillation: what MDH names for uranium, and why you cannot verify it

The bottom line

Find the radionuclide lines on your report before doing anything else. Exceedance is uncommon and concentrated in groundwater systems, and EPA estimated only 795 systems nationally would need to install treatment. If your figure is under the limit, no filter purchase follows. If radium is the number that concerns you, certified products exist on two routes: 114 point-of-entry softeners and 120 point-of-use reverse osmosis systems. If uranium is the number, no certified household claim exists at all, and that is worth knowing before anyone sells you one.

Who this is not for

This is about radionuclides on a public water system, where a utility monitors and reports them. It does not cover radon in indoor air, which is a different and generally larger exposure with its own testing route. It does not cover private wells, where nobody monitors anything unless the owner pays. It names no model and is not medical advice on radiation exposure.

Questions

What is the limit for radium in drinking water?

EPA sets combined radium-226 and radium-228 at 5 picocuries per litre. The maximum contaminant level goal is zero, which is a non-enforceable health goal rather than the standard. The enforceable figure your utility must meet is the 5 pCi/L maximum contaminant level.

What is the uranium limit in tap water?

EPA sets uranium at 30 micrograms per litre, a standard issued under the Safe Drinking Water Act Amendments of 1986 and finalised in 2000. EPA names increased cancer risk and kidney toxicity as the health effects of long-term exposure above it.

Can a water filter remove uranium?

No certified one. NSF offers no selectable uranium reduction claim under any standard, so no product carries a listed uranium performance. That is a gap in certification, not proof no technology works — EPA and state health departments both name treatments for uranium.

Does a water softener reduce radium?

Some are certified to. NSF listings show 114 softener models certified for radium 226/228 reduction under NSF/ANSI 44, all point-of-entry. But 78 of the 192 hardness-certified models hold no radium claim, so softening certification and radium certification are not the same thing.

Is radon in drinking water regulated?

No. EPA states the gross alpha standard covers all alphas but explicitly does not include radon and uranium. A radon rule was proposed and remains proposed, so a utility has no enforceable radon figure in water to meet or report.

How old is the radionuclide number on my report?

Possibly several years. After four quarters of initial monitoring, EPA reduces the frequency to one sample every nine years where results were below the detection limit, every six years at or below half the limit, and every three years nearer to it.

How common is a radium exceedance?

Uncommon, and concentrated in groundwater. Minnesota Department of Health reports that in 2014 nine of Minnesota's municipal systems — 1.2 percent — exceeded the radium standard, and states surface water generally does not contain radionuclides at levels of concern.

Where do radium and uranium come from?

EPA names erosion of natural deposits as the source of both. They are not industrial contaminants in the usual sense, which is why occurrence tracks geology: Minnesota's highest levels sit in the Mount Simon-Hinckley and Jordan aquifers in the southeast of the state.

Sources — 14 claims

EPA — Radionuclides Rule: A Quick Reference Guide, EPA 816-F-01-003 (June 2001)
EPA sets four radionuclide standards in drinking water: gross alpha particle activity at 15 pCi/L, combined radium-226 and radium-228 at 5 pCi/L, beta particle and photon emitters at 4 millirems per year, and uranium at 30 micrograms per litre. The maximum contaminant level goal for every one of the four is zero. (captured 2026-09-11)
Limits: An MCLG of zero is a non-enforceable health goal, not the standard. The enforceable figure is the MCL. Minnesota Department of Health states the uranium standard as 20 pCi/L, which matches EPA neither in value nor in unit; EPA's own quick reference guide and its National Primary Drinking Water Regulations table both give 30 µg/L, and that figure is used here.
EPA — Radionuclides Rule: A Quick Reference Guide, EPA 816-F-01-003 (June 2001)
EPA states the gross alpha particle standard of 15 pCi/L covers all alphas but does not include radon and uranium, and that a total of 168 individual beta particle and photon emitters may be used to calculate compliance with the 4 mrem per year standard. (captured 2026-09-11)
Limits: Radon in drinking water has no enforceable federal standard at all. EPA published a proposed radon rule, and it remains proposed, so a utility has nothing to comply with on radon.
EPA — National Primary Drinking Water Regulations, radionuclides table
EPA states the source of radium-226 and radium-228 and of uranium in drinking water is erosion of natural deposits, that the health effect of long-term exposure above the MCL for each of the four regulated radionuclides is increased risk of cancer, and that uranium also carries kidney toxicity. (captured 2026-09-11)
Limits: Health effects stated for long-term exposure above the MCL. EPA gives no figure here for risk at any particular concentration below it.
Minnesota Department of Health — Radionuclides (Radium) in Drinking Water (page last updated 2024-09-25)
Minnesota Department of Health states radionuclides occur naturally in Minnesota groundwater, that surface water generally does not contain them at levels of concern, and that in 2014 nine of Minnesota's municipal water systems — 1.2 percent — had radium in treated water above the EPA standard, with the highest source-water levels in the Mount Simon-Hinckley and Jordan Aquifers in southeastern Minnesota. (captured 2026-09-11)
Limits: One state, one year. It establishes that exceedance is uncommon and geographically concentrated in groundwater systems, not that any figure applies elsewhere. A system on surface water is a different case.
EPA — Radionuclides Rule: A Quick Reference Guide, EPA 816-F-01-003 (June 2001)
EPA's Radionuclides Rule, published 7 December 2000, retained the existing standards for combined radium-226/228, gross alpha and beta/photon activity and regulated uranium for the first time. It applies to community water systems of all size categories, and EPA estimated only 795 systems would have to install treatment, at an annual compliance cost of 81 million dollars. (captured 2026-09-11)
Limits: An estimate made in 2001 of how many systems the rule would force into treatment. It is not a current count of systems out of compliance, and it says nothing about any particular utility.
EPA — Radionuclides Rule: A Quick Reference Guide, EPA 816-F-01-003 (June 2001)
EPA requires four consecutive quarters of initial radionuclide monitoring at each entry point to the distribution system, then reduces the frequency by result: one sample every nine years where the average was below the detection limit, every six years where it was at or below half the MCL, and every three years where it was above half the MCL but at or below the MCL. (captured 2026-09-11)
Limits: The reduced schedule is why a radionuclide figure on an annual report can be several years old, up to nine. A system whose entry point result exceeds the MCL must return to quarterly sampling until four consecutive samples are below it.
EPA — Radionuclides Rule (drinking water regulation information page)
EPA states it issued the uranium standard as required by the Safe Drinking Water Act Amendments of 1986, and that the Radionuclides Rule set new monitoring provisions for community water systems to ensure all their customers receive water meeting the radionuclide maximum contaminant levels. (captured 2026-09-11)
Limits: Fourteen years passed between the statutory mandate and the standard. The gap is a matter of record and this page draws no conclusion from it.
Minnesota Department of Health — Radionuclides (Radium) in Drinking Water (page last updated 2024-09-25)
Minnesota Department of Health states that community public water systems test for radionuclides and ensure levels meet EPA standards, that a household can find the levels its system detected by reading the Consumer Confidence Report, and that home treatment systems such as ion exchange water softeners and reverse osmosis systems can reduce the levels of some radionuclides. (captured 2026-09-11)
Limits: MDH says some radionuclides, not all. NSF offers a certified reduction claim for radium 226/228 only; there is no certified claim for uranium, gross alpha or beta/photon emitters under any NSF standard.
NSF Official Listings — Drinking Water Treatment Units, NSF/ANSI 44 claim searches (listings current as of Thursday, September 10, 2026, 12:15 a.m. Eastern Time)
NSF listings returned 115 rows for radium 226/228 reduction under NSF/ANSI 44 on 2026-09-10, resolving to 114 distinct softener models from 8 companies. Every one is listed point-of-entry. (captured 2026-09-11)
Limits: A softener claim under the cation exchange standard. It is a different certification from the NSF/ANSI 58 radium claim held by reverse osmosis systems, and the two are structurally opposite: these are whole-house, those are point-of-use.
NSF Official Listings — Drinking Water Treatment Units, NSF/ANSI 44 claim searches (listings current as of Thursday, September 10, 2026, 12:15 a.m. Eastern Time)
Under NSF/ANSI 44 the radium 226/228 and barium claim sets are the same 114 models on 2026-09-10, and all 114 sit inside the 192 models certified for hardness reduction. 78 hardness-certified softeners hold neither claim, so a softener certified to soften is not thereby certified to reduce radium. (captured 2026-09-11)
Limits: A statement about which models carry which listed claim. It does not say an uncertified softener fails to reduce radium — ion exchange is the mechanism either way — only that its performance has not been tested and listed.
NSF Official Listings — Drinking Water Treatment Units, NSF/ANSI 44 claim searches (listings current as of Thursday, September 10, 2026, 12:15 a.m. Eastern Time)
Radium 226/228 is the only radionuclide NSF offers a reduction claim for, and it appears under exactly two standards: NSF/ANSI 44 with 114 certified softener models and NSF/ANSI 58 with 120 certified reverse osmosis models. No claim for uranium, gross alpha or beta and photon emitters is selectable under any NSF standard, so three of EPA's four regulated radionuclide entries have no certified household product at all. (captured 2026-09-11)
Limits: Established by searching NSF's full selectable claim list for uranium and finding no option, with radium as the positive control: radium returns exactly two options, so the search works and the uranium absence is real. NSF listings only — WQA and IAPMO certify separately. The absence of a claim is not evidence that no technology works; EPA and MDH both name treatments for uranium.
NSF Official Listings — Drinking Water Treatment Units, reduction-claim search form (captured 2026-09-10)
NSF offers four selectable reduction claims under NSF/ANSI 44, the cation exchange softener standard: barium reduction, efficiency rated, hardness reduction and radium 226/228 reduction. No nitrate claim appears under it. (captured 2026-09-10)
Limits: The claims selectable in NSF search on the capture date. Ion exchange is a technology EPA names for nitrate, so the absence of a nitrate claim here bounds what a softener can be certified for rather than what ion exchange can do.
NSF Official Listings — Drinking Water Treatment Units, NSF/ANSI 58 TDS reduction claim search (listings current as of Thursday, September 10, 2026, 12:15 a.m. Eastern Time)
The optional NSF/ANSI 58 claims are structured as one bundle with two nested tiers on 2026-09-10. Seven claims are held by exactly the same 127 models — lead, fluoride, barium, cadmium, trivalent chromium, copper and selenium. Hexavalent chromium and radium 226/228 are held by the same 120 models, all inside that 127. Nitrate/nitrite is held by 84, all inside that 120. Turbidity (61) and asbestos (10) also sit wholly inside the 127. (captured 2026-09-11)
Limits: A structural observation about how certification is bought, not about membrane chemistry. Manufacturers evidently test to a standard package rather than claim by claim, so the claim a model holds tells you which package was purchased.
Crosswalk of EPA National Primary Drinking Water Regulations against the NSF reduction-claim taxonomy (both captured 2026-09-10)
Crossing EPA 94 regulated contaminant entries against NSF 101 selectable reduction claims on 2026-09-10 leaves 33 regulated contaminants with no NSF reduction claim of that name: 19 organic chemicals, 4 microorganisms, 4 inorganic chemicals, 3 disinfection byproducts and 3 radionuclides. (captured 2026-09-10)
Limits: The absence of a claim by name is not evidence that no product treats the contaminant. Reverse osmosis and distillation remove many of these; NSF offers no named claim to certify them against. NSF offers no claims at all under its distillation standard, so distillation is invisible to this crosswalk, and microorganisms are certified through NSF/ANSI 55 disinfection performance classes rather than by contaminant.