Radionuclides (uranium, radium)

WTR

20 µg/L EPA's own health-derived level for uranium (micrograms per liter)

EPA derived 20 from kidney toxicity, judged it feasible, then set the enforceable limit at 30 on cost. Radium and gross alpha have no health line to grade against, only goals a lab cannot measure, so their legal limits do that job.

Contested, tiered honestly

Stasis Home Research Every claim on this page is graded against peer-reviewed health research and linked to its primary source.Updated August 2026

Radionuclides on a water report are three different measurements, and they are almost entirely natural. Radium is a bone-seeking carcinogen. Uranium is a kidney concern that no agency has classified as a carcinogen, and gross alpha is a screening count rather than a substance. Reverse osmosis at the kitchen tap removes uranium and radium well. Carbon does not touch either one.

Radionuclides (uranium, radium): what is it, and where does it come from?

The word covers three things that answer three different questions, and running them together is the most common error in consumer water writing. Combined radium is a specific measurement of two isotopes, radium-226 and radium-228, and the federal limit applies to the two added together. Uranium is a specific element, measured either by mass in micrograms per liter or by activity in picocuries per liter. Gross alpha particle activity is not a substance at all: it is a count of total alpha emissions per liter from whatever mixture happens to be in that water, cheap and fast, which is why utilities measure it first and then go looking for the specific isotopes behind a high result. The federal definition of the regulated gross alpha figure includes radium-226 and excludes both radon and uranium.

All three come from geology, not from industry. Uranium is present in nearly all rocks and soils and dissolves into groundwater over long rock-contact timescales, and radium-226 and radium-228 are decay products of the same uranium and thorium in the same rock. Imported surface water spends too little time against rock to pick up much of either, so the radiological signal lives in groundwater-reliant systems. Nobody put it there, and it is still worth removing.

One trap is worth naming before you read any number on this page, because two of them look identical and are not. Uranium is the only parameter here measured in two incompatible quantities. Twenty micrograms per liter is EPA's health-derived level, a mass. Twenty picocuries per liter is California's enforceable limit, an activity. Same digits, different quantity, different authority, and converting between them requires knowing the isotope mix of your specific water, which no monitoring file records. So we never convert a reading. Whatever unit your utility reported, we read it against the limit that authority wrote in that same unit.

Radionuclides (uranium, radium): why does it matter?

Radium and uranium harm the body by different routes, and the honest strength of the evidence is different for each.

Radium is the radiological cancer parameter. It sits with calcium in the periodic table, so the body handles ingested radium as a calcium analogue and deposits it in bone, where its alpha decays irradiate bone surfaces and nearby marrow at close range for years. The human evidence comes from one tragic occupational cohort: US watch-dial painters who shaped brush tips with their lips in the 1910s and 1920s. IARC classifies radium-224, radium-226 and radium-228 as Group 1, carcinogenic to humans, with bone cancer as the site of sufficient evidence. Here is the constraint that goes with that evidence, and both halves of it matter. Those women carried body burdens orders of magnitude above anything drinking water produces, and the 5 pCi/L limit comes from extrapolating their risk downward on a no-threshold model, not from observing harm at 5 pCi/L. A household at 6 pCi/L is not on the dial painters' curve. It still is not good, and a bone-seeking Group 1 carcinogen accumulating for years, in a body that is also carrying whatever else is in that water, is worth removing.

Uranium is the parameter where precision matters most, because two credible agencies say things that sound contradictory and are not. Its established health concern in drinking water is the kidney, specifically the proximal tubule, and water-soluble uranium compounds, which is what drinking water carries, reach the kidney at lower doses than insoluble ones. Separately, EPA treats uranium's radioactivity as carrying a no-threshold cancer risk, which is why the federal health goal is zero and why California's health goal was calculated on a cancer-risk basis. And yet no agency has classified uranium itself as a human carcinogen.

That pair looks like a dodge and it is not, so here is the difference. A cancer classification is a verdict on evidence in people and animals: IARC, NTP and EPA look for studies showing that this substance caused cancer at doses people actually received, and for natural uranium those studies do not exist in sufficient quantity or quality, so no agency has issued a verdict either way. Not classified means the question is open, not that the answer came back negative. A health goal of zero is not a verdict at all. It is the output of a precautionary rule EPA applies to every source of ionizing radiation, in the agency's own words: because uranium is radioactive, and because EPA models radiation risk as having no safe threshold, the goal is set at zero by construction. The same rule sets the goal at zero for combined radium and for gross alpha too. A reader who has understood that pair understands more about how drinking-water limits are built than most consumer water content will ever tell them.

How much a low-level reading means is genuinely contested, and we say so rather than picking the scarier reading and hiding the other. The best human drinking-water study, a Finnish drilled-well cohort, found uranium exposure weakly associated with altered proximal tubule function without a clear threshold. WHO reads the same literature more conservatively and holds that there is no clear evidence of effects below 30 µg/L, on the ground that the observed changes stayed within the normal physiological range. The distinction to keep is between a measurable change in how the kidney handles calcium and phosphate, which the studies do show at drinking-water levels, and kidney disease, which they do not.

Gross alpha has no mechanism of its own, because it is not a substance. Its health meaning is entirely borrowed from whatever emitters produce the count. A high gross alpha with no isotope breakdown is a gap in the record rather than a finding about your water, and the next step is a test that speciates it.

Per ATSDR, the health effects of natural and depleted uranium come from its chemical action rather than its radiation, the kidney is the target organ, water-soluble uranium compounds cause kidney effects at lower doses than insoluble ones, and neither NTP, IARC nor EPA has classified natural or depleted uranium as to carcinogenicity.ATSDR, Natural and Depleted Uranium ToxFAQs (CAS 7440-61-1, February 2013). Health effects due to chemical rather than radiological action; kidney the target organ; soluble compounds act at lower doses; not classified as to carcinogenicity by NTP, IARC or EPA
In the 2000 Radionuclides Rule, EPA set the uranium health goal at zero because it models ionizing radiation risk as having no safe threshold, derived a health-based drinking water level of 20 µg/L from kidney toxicity, judged that level feasible, and then set the enforceable limit at 30 µg/L on the finding that the benefits do not justify the costs at the feasible level.US EPA, National Primary Drinking Water Regulations: Radionuclides, Final Rule, 65 FR 76708 (December 7, 2000). Uranium health goal zero under the non-threshold linear radiation risk model; health-derived level of 20 µg/L from kidney toxicity; enforceable limit set at 30 µg/L because the benefits do not justify the costs at the feasible level; mass-to-activity best estimate 0.9 pCi/µg with an observed range of 0.67 to 1.5
Among 1,474 women employed in the US radium dial-painting industry before 1930, there are 61 known cases of bone sarcoma and 21 cases of carcinoma of the paranasal sinuses or mastoid air cells.Rowland, Stehney and Lucas, 1978 (Radiation Research 76(2):368-383), dose-response relationships for female radium dial workers: 61 bone sarcomas and 21 head carcinomas among 1,474 women employed in the US radium dial-painting industry before 1930
IARC classifies radium-224, radium-226 and radium-228 and their decay products as Group 1, carcinogenic to humans, with bone cancer the site of sufficient evidence in humans.IARC Monographs Volume 100D, Radiation (2012). Radium-224, radium-226 and radium-228 and their decay products classified Group 1, carcinogenic to humans, with sufficient evidence in humans for bone cancer
A 2025 systematic review of 29 studies of radium, uranium and radon in drinking water concluded that the evidence base cannot support a pooled cancer risk estimate: most studies are ecological rather than individual-level, they are too heterogeneous to combine, and polonium and radioactive lead are generally unaddressed. This is why we treat the regulatory limits as downward extrapolations rather than as observed harm at those levels.Alimam and Auvinen, 2025 (Science of the Total Environment 968:178849), systematic review of cancer risk from naturally occurring radionuclides in drinking water: 29 studies of radium, uranium and radon, mostly ecological rather than individual-level and too heterogeneous for a pooled risk estimate
In 325 Finnish drilled-well users at a median water uranium of 28 µg/L, urinary uranium was significantly associated with increased fractional excretion of calcium and phosphate but not with creatinine clearance or urinary albumin, which the authors read as a weak association with altered proximal tubule function without a clear threshold.Kurttio et al., 2002 (Environmental Health Perspectives 110(4):337-342), renal effects of uranium in drinking water: 325 drilled-well users at a median water uranium of 28 µg/L; increased fractional excretion of calcium and phosphate; no association with creatinine clearance or urinary albumin; a weak association with altered proximal tubule function without a clear threshold
WHO weighs the same literature more conservatively: its 30 µg/L guideline value is designated provisional because of scientific uncertainties surrounding uranium toxicity, nephritis is named the primary chemically induced effect in humans, the carcinogenicity data are called insufficient, and WHO finds no clear evidence of effects below 30 µg/L.WHO, Uranium in drinking-water, chemical fact sheet, Guidelines for Drinking-water Quality. Provisional guideline value of 30 µg/L, designated provisional because of scientific uncertainties surrounding uranium toxicity; nephritis the primary chemically induced effect in humans; insufficient data on carcinogenicity; no clear evidence of effects below 30 µg/L; only chemical aspects addressed
Across 37,915 US community water systems in EPA's 2006 to 2011 monitoring records, uranium had the highest share of records with detections at 63.1 percent, and 2.1 percent of analysed systems had average concentrations above the federal limit. The authors also found metal concentrations, uranium among them, elevated in systems serving semi-urban Hispanic communities independent of location or region, and it is that independence from region that led them to attribute the disparity to regulatory policy or treatment rather than to geology. The study did not analyse radium and this finding must not be extended to it.Ravalli et al., 2022 (The Lancet Planetary Health 6(4):e320-e330), sociodemographic inequalities in uranium and other metals in US community water systems, 2006 to 2011: 37,915 systems; uranium the highest share of records with detections at 63.1 percent; 2.1 percent of analysed systems above the limit; metal concentrations elevated in systems serving semi-urban Hispanic communities independent of location or region. The study analyses ten metals and metalloids and does not analyse radium
A 2025 peer-reviewed review of removal technologies puts membrane processes at 90 to 99 percent for both radium and uranium under laboratory conditions.Al-Shomali et al., 2025 (International Journal of Environmental Research and Public Health 22(5):727), state-of-the-art review on removal of naturally occurring radioactive materials in water: membrane processes 90 to 99 percent for both radium and uranium; lime softening 80 to 95 and 85 to 99; coagulation and filtration 65 to 95 and 80 to 95
Field testing of household units lands lower than the laboratory range: Nova Scotia's environment department reports that a reverse osmosis unit generally removes 90 percent of the uranium, with efficiency depending on the manufacturer and the type of membrane.Nova Scotia Environment, Removing Uranium From Drinking Water Using Reverse Osmosis: tests show that the reverse osmosis unit generally removes 90 percent of the uranium, with efficiency depending on the manufacturer and the type of membrane

Radionuclides (uranium, radium): what do we grade it against?

Radionuclides (uranium, radium) is held to two different numbers, and they do not agree. The health-based level for Uranium by activity, California (pCi/L) is 0.43 (California's public health goal, set on a one-in-a-million lifetime cancer risk from ionizing radiation and corroborated by a kidney no-observed-effect level. It is a goal, not a flagging line: it sits below the 1 pCi/L level a compliance lab reports, so reading against it would flag essentially every detection), from California OEHHA public health goal for uranium, 2001, via the CA SWRCB MCL/DLR/PHG table (September 2025). The enforceable legal limit is 20. The table below lists every level on this page next to the source that set it.

Where the health research draws the line, versus the legal limit.
Contaminant	Health-based level	Legal limit	Source (health-based)
Uranium by activity, California (pCi/L)	0.43 California's public health goal, set on a one-in-a-million lifetime cancer risk from ionizing radiation and corroborated by a kidney no-observed-effect level. It is a goal, not a flagging line: it sits below the 1 pCi/L level a compliance lab reports, so reading against it would flag essentially every detection	20 California's enforceable limit, 22 CCR 64442, effective 2006-06-11. There is no federal uranium limit expressed in picocuries per liter anywhere in the CFR, so this is the applicable line for a reading in this unit	California OEHHA public health goal for uranium, 2001, via the CA SWRCB MCL/DLR/PHG table (September 2025)
Uranium by mass, federal (µg/L)	20 EPA's own health-derived drinking water level from kidney toxicity, in the 2000 Radionuclides Rule. EPA judged 20 feasible and then set the enforceable limit at 30 under its cost-benefit authority, finding that the benefits do not justify the costs at the feasible level. This is the line a microgram reading gets read against. California's goal in this same unit is 0.5 µg/L, which is OEHHA's own published pair for the 0.43 pCi/L above, not a conversion of ours	30 the federal limit, 40 CFR 141.66(e), with compliance required from 2003-12-08	US EPA, Radionuclides Final Rule, 65 FR 76708 (December 7, 2000)
Combined radium-226 and radium-228 (pCi/L)	0.069 our own sum of California's two per-isotope public health goals, radium-226 at 0.05 and radium-228 at 0.019. California publishes no combined goal, so the addition is ours, and we label it as ours. It sits about 14 times below the 1 pCi/L a lab reports, so nothing is graded against it	5 the federal limit, 40 CFR 141.66(b), and it applies to the two isotopes added together	California OEHHA public health goals for radium-226 and radium-228, 2006
Gross alpha particle activity (pCi/L)	no health-based number exists. The federal health goal is zero, and California's own risk office concluded in 2003 that it would not be practical to develop one, because gross alpha is a screening sum of an unknown mixture rather than a substance. The 15 pCi/L limit is what a reading is graded against	15 the federal limit, 40 CFR 141.66(c), for the regulated form that includes radium-226 and excludes radon and uranium. A separate, unregulated screening figure that adds uranium back in is systematically higher and has no limit at all	40 CFR 141.55 (health goal zero); CA SWRCB MCL/DLR/PHG table, footnote 8 (no gross alpha public health goal)

Read the unit in each row label before comparing anything. Uranium is measured in two incompatible quantities, converting between them requires knowing the isotope mix of your specific water, and we never convert a reading. Two of these numbers read 20 and they are different things: 20 micrograms per liter is EPA's health-derived level, 20 picocuries per liter is California's enforceable limit. So uranium is graded two different ways: a microgram reading against EPA's health-derived 20, and a picocurie reading against California's 20 pCi/L limit, because no health line exists in that unit at all. Combined radium and gross alpha are graded against their legal limits, because neither carries a health number that can act as a line: radium's 0.069 is our own sum of California's two per-isotope goals and sits about fourteen times below what a lab can report, and gross alpha has none beyond the federal goal of zero. Gross beta activity is a separate parameter this page does not grade.

What helps against Radionuclides (uranium, radium)?

5 approaches below reduce Radionuclides (uranium, radium) in a home, ordered from the most direct fix to the smallest change in habit. The most direct is Reverse osmosis at the kitchen tap (NSF/ANSI 58). Each entry also names what it cannot do, because no single fix covers every exposure route.

Direct fixes

Reverse osmosis at the kitchen tap (NSF/ANSI 58)
The only technology EPA designates as best available for all three parameters, which is why the residential answer collapses to one product. Laboratory membrane performance runs 90 to 99 percent for both uranium and radium; a field programme testing household units reports about 90 percent for uranium, so a real under-sink unit does around 90 percent or better. Where a gross alpha count is driven by radium and uranium, removing those two controls the count. It is the same under-sink unit already recommended for PFAS, chromium-6 and nitrate.
It treats one tap, which is the right scope here, because the exposure route for all three is drinking rather than showering or breathing. Membrane capacity is finite and breakthrough is invisible: no taste change, no odour, no flow change. Replace on a schedule, not on a symptom.

Bigger retrofits

Ion exchange at the point of entry or at utility scale
Works, and works differently for each. Radium is a divalent cation, so cation exchange removes it, which is the same chemistry as a conventional water softener. Uranium in oxygenated water travels as a negatively charged uranyl-carbonate complex, so it needs anion exchange, which a softener does not provide.
That asymmetry is a real trap: if you run a softener you are partly covered for radium and not at all for uranium. Whole-house ion exchange also concentrates a radioactive residual in the resin and brine, which is a disposal question no household should be handed casually. Point-of-use reverse osmosis avoids it.

Free and behavioral

Activated carbon
Useful for chlorination byproducts and taste, and no help at all here.
Activated carbon does not remove uranium or radium. A household with a detected radionuclide and only a carbon block or pitcher needs reverse osmosis, the same upgrade point that applies to arsenic and chromium-6. There is no gross alpha filter either, because gross alpha is a count rather than a substance.
Do not boil it off
Boiling is the common reflex and it is the wrong move for all three. Evaporation removes water and leaves the radionuclides behind, so it concentrates them rather than removing them.
If gross alpha is elevated with no isotope breakdown, test for the isotopes
A high screening count is a prompt to go look, not a finding in itself. Utilities are allowed to substitute a low gross alpha result for the specific radium and uranium tests, so a low screen is a legitimate answer and a high one leaves the actual question open. A tap test that reports radium-226, radium-228 and uranium separately closes the gap, and it tells you which of the two treatment routes you actually need.

See how well each fix works

There is no villain in this one and we are not going to invent one. Uranium and its radium decay products were in the aquifer before anyone drilled a well into it, and an industry-versus-regulator story here would be false. Naturally occurring describes the origin, not the consequence: nobody's fault, still worth removing. The reasoning we owe you is the same reasoning the regulator used: the federal health goal for all three is zero, the enforceable lines are not, and in uranium's case EPA wrote down why in its own final rule, having judged 20 micrograms per liter feasible and then set the limit at 30 because the benefits did not justify the costs at the feasible level. That is a cost-benefit judgment sitting on top of a health judgment, stated openly, and it is a legitimate thing to show you. What it costs to close at your own tap is one under-sink unit you may already be installing for something else.

Common questions about Radionuclides (uranium, radium)

Is uranium in tap water a cancer risk?

No agency has classified uranium as a carcinogen, and its established health concern in drinking water is the kidney rather than cancer. That sits next to a fact that sounds like a contradiction: the federal health goal for uranium is zero, and EPA says the reason is cancer risk from its radioactivity. Both are true, because they answer different questions. A classification is a verdict on evidence in people and animals, and for natural uranium those studies do not exist in sufficient quantity or quality, so no agency has ruled either way. Not classified means the question is open, not that the answer came back negative. A goal of zero is not a verdict at all, it is a precautionary rule EPA applies to everything radioactive, on a model in which radiation risk has no safe threshold. Radium, a different parameter on the same report, is the one with a classified cancer mechanism.

What is gross alpha on my water report?

A screening measurement, not a substance you have. It counts total alpha-particle emissions per liter from whatever mixture is in the water, and the regulated form includes radium-226 while excluding radon and uranium. Utilities measure it first because it is cheap, and the regulations let a low result stand in for the specific radium and uranium tests. So a low gross alpha is a legitimate answer to whether radium or uranium is present, and a high one is a prompt to go look at the specific isotopes rather than a finding on its own. There is no gross alpha filter, because you cannot filter a count. Controlling the emitters controls the number.

Is the radium in my water the same thing that harmed the radium dial painters?

It is the same element and it is not the same exposure, and both halves of that matter. The dial painters ingested radium by shaping brush tips with their lips and carried body burdens orders of magnitude above anything drinking water produces, which is how we know radium causes bone cancer in humans at all. The 5 pCi/L limit was not set by observing harm at 5 pCi/L; it comes from extrapolating that occupational risk downward on a model with no safe threshold. A household at 6 pCi/L is not on their curve. At the same time, the epidemiology at ordinary drinking-water levels is thin: a 2025 systematic review of 29 studies found them too heterogeneous and too often ecological to support a pooled risk estimate. A bone-seeking Group 1 carcinogen that accumulates for years is still worth removing, and the reason to remove it does not need the dial painters' numbers attached.

My uranium is reported in µg/L but the limit I see is in pCi/L. How do I compare them?

You do not, and neither do we. Micrograms per liter is a mass and picocuries per liter is a rate of radioactive decay, and converting between them requires knowing the isotope mix of your specific water, which no monitoring file records. Three official conversion factors exist, EPA states the range observed in drinking water as 0.67 to 1.5, and in the water records we hold, across the systems that reported both quantities, the implied factor scatters about fifteen-fold. So any single factor applied to a single sample produces a number nobody could defend on your panel. What we do instead is read each reading against the limit that authority wrote in that same unit: a microgram reading against the federal 30 µg/L limit and EPA's health-derived 20 µg/L, a picocurie reading against California's 20 pCi/L. If you want the rough equivalence for orientation, use EPA's own sentence: assuming a conversion factor of 0.9, an MCL of 30 µg/L will typically correspond to 27 pCi/L. That is a rough range, not arithmetic anyone should apply to your number.

Where do radionuclides in drinking water come from?

Geology, essentially always. Uranium is present in nearly all rocks and soils, and the longer groundwater sits against that rock, the more of it dissolves in. Radium-226 and radium-228 form as the uranium and thorium in the same rock decay. Water imported from surface sources spends too little time against rock to pick up much of either, so the signal lives in groundwater-reliant systems, and where you land depends on which aquifer you are drinking from rather than on anything a utility did. Human activity can add to the natural background through uranium mining and milling, coal combustion, and phosphate fertilizer manufacture, but it is an increment on a natural source rather than the source itself.

Will a carbon filter or boiling remove uranium or radium?

Neither one. Activated carbon does not remove uranium or radium, so a carbon pitcher or under-sink carbon block is the wrong tool here. Boiling is worse than useless: it evaporates water and leaves the radionuclides behind, so it concentrates them. The point-of-use route that works is reverse osmosis, which EPA designates as a best available technology for all three parameters and which household field testing puts at around 90 percent removal for uranium. Ion exchange handles them at whole-house or utility scale, with the catch that a water softener covers radium partly and uranium not at all.

My water is under the legal limit. Does that mean it is fine?

It means your utility is meeting the law, which is worth something and is not the same question as what is in your glass. The federal health goal for all three of these parameters is zero, and the enforceable limits are not. For uranium the gap is written down in EPA's own final rule: the agency derived 20 micrograms per liter from kidney toxicity, judged that level feasible, and then set the enforceable limit at 30 because it found the benefits did not justify the costs at the feasible level. Combined radium and gross alpha work differently. Gross alpha has no health-based number beyond that goal of zero, and the only one that exists for combined radium is our own sum of California's two per-isotope goals, 0.069 picocuries per liter, which sits about fourteen times below what a compliance lab can report. California's uranium goal of 0.43 picocuries per liter sits below that reporting floor the same way. Those are goals, not lines anything gets flagged against, which is why we read radium and gross alpha against their legal limits. Reverse osmosis at the kitchen tap closes the distance without changing anything about your water service.