Home › Guides › Pharmaceuticals in Drinking Water

Pharmaceuticals in Drinking Water: What the Data Actually Shows

Pharmaceutical compounds in drinking water is a subject where the headline numbers and the sampling data disagree badly. The most-quoted statistic in the field — that pharmaceuticals turn up in 80% of American water — describes streams, not tap water, and includes compounds that aren't pharmaceuticals at all. Meanwhile the genuinely significant fact gets almost no coverage: EPA has never set an enforceable limit on a single pharmaceutical compound, and in April 2026 it proposed listing pharmaceuticals as a regulated-candidate chemical group for the first time. This guide separates the sampling data from the marketing, explains what conventional treatment does and doesn't remove, and covers what actually works.

What has actually been detected

Analytical chemistry got dramatically better in the 2000s. Liquid chromatography with tandem mass spectrometry made it routine to quantify organic compounds at nanograms per liter — parts per trillion. A great deal of what reads as "contamination is getting worse" is really detection getting better. Compounds that were present all along at 5 ng/L simply weren't measurable before.

What that instrumentation found, across USGS, EPA, and academic sampling programs, spans essentially every therapeutic class in common use:

• Analgesics and anti-inflammatories — ibuprofen, acetaminophen, naproxen. Among the most frequently detected, for the obvious reason that they're the most heavily consumed.
• Antibiotics — sulfamethoxazole, trimethoprim, erythromycin. These carry a distinct concern beyond direct human exposure: environmental selection pressure for antibiotic resistance.
• Anticonvulsants — carbamazepine and phenytoin. Carbamazepine in particular is chemically stable, poorly biodegradable, and resistant to conventional oxidation, which makes it a standard tracer compound for wastewater influence on a water source.
• Hormones — estrone, estradiol, and synthetic analogs. Detected less often than the drug classes above, but with lower effect thresholds, which is why they get research attention out of proportion to their detection frequency.
• Anxiolytics and cardiovascular drugs — meprobamate, atenolol, gemfibrozil.
• Stimulants and lifestyle compounds — caffeine, cotinine, DEET. Not pharmaceuticals in the therapeutic sense, but routinely reported alongside them as wastewater indicators, which is a large part of why detection statistics get inflated.

The 80% number, and what it really measured

Nearly every consumer article on this subject cites a USGS finding that roughly 80% of tested waterways contained pharmaceutical compounds, and most present it as though it describes drinking water. It doesn't.

The drinking-water-source equivalent of that study exists, and it tells a different story. USGS analyzed 103 pharmaceuticals and 21 hormones across nearly 1,100 wells tapping aquifers that supply drinking water to roughly 80 million people — aquifers representing about 60% of the groundwater volume pumped for drinking supply nationally. Samples were collected before any treatment.

Detection frequencies and concentrations in groundwater came in lower than what other studies have reported for surface water. When compounds were found, they were at low concentrations not expected to produce adverse human health effects.

Source water vs. finished water

For anyone deciding what treatment to install, the useful question isn't what's in the river — it's what makes it through the treatment plant and out of the tap. EPA and USGS ran a joint multi-phase study specifically to answer that, sampling source and treated water at approximately 25 drinking water treatment plants for 247 chemical and microbiological pollutants. Plants were chosen for exposure to municipal waste, septic systems, and livestock production — a deliberately worst-case selection, not a representative national average.

A companion study across 19 treatment plants gives the clearest picture of what survives treatment. Eleven compounds appeared in more than half of source water samples. Only three appeared in more than half of finished or distribution system samples:

The pattern is consistent and mechanistically sensible. Conventional treatment — coagulation, sedimentation, sand filtration, chlorination — was engineered to remove turbidity and inactivate pathogens. It does that well. It was never designed to remove dissolved organic micropollutants, and published research confirms that coagulation and chlorination are ineffective against pharmaceuticals as a class. The compounds that survive share a profile: polar, water-soluble, chemically stable, and resistant to oxidation. The ones that don't survive are generally the ones chlorine can oxidize or that adsorb onto floc.

The regulatory gap — and CCL 6

Here is the single most important fact in this subject, and the one that vendor content mentions only in passing before pivoting to a product: EPA has never established a Maximum Contaminant Level for any pharmaceutical compound. Not one. There is no enforceable federal limit, no required monitoring, and no violation threshold. A utility can deliver water containing carbamazepine at any concentration and remain in complete compliance with the Safe Drinking Water Act, because no standard exists to violate.

That's not regulatory failure so much as regulatory sequence. Under the SDWA, EPA works through a defined pipeline, and pharmaceuticals only just entered it as a group.

What the CCL is not, and this distinction gets lost constantly: the CCL imposes no requirements on public water systems whatsoever. It's a prioritization and pipeline mechanism. Its practical consequences are that it feeds candidate selection for the Unregulated Contaminant Monitoring Rule — the rule requiring systems to monitor for up to 30 unregulated contaminants to fill occurrence data gaps — and that it obligates EPA to eventually make a regulatory determination on each listing.

NSF 401 and the certification blind spot

Because there's no MCL, there's no compliance-driven certification either. The only common product standard that tests specifically for pharmaceutical compounds is voluntary — and most buyers have never heard of it.

The practical consequence: a filter marketed with imagery about "emerging contaminants" while carrying only NSF 42 certification has been tested for chlorine taste and nothing else. That's not necessarily deceptive — the box may make no explicit pharmaceutical claim — but the buyer's inference is wrong, and the certification logo is doing work it wasn't designed to do. The full standard-by-standard breakdown is in the NSF water filter certifications guide, and the engineering side of why budget systems underperform is covered in the budget filters vs. professional systems comparison.

What actually removes these compounds

The technology picture is well-documented and, unusually for this field, fairly encouraging. Advanced treatment methods work. Published research shows ozonation, granular activated carbon, reverse osmosis, and nanofiltration can each remove more than 99% of targeted pharmaceuticals.

A frequently cited Korean study is a useful illustration. Six micropollutants were detected at measurable levels in Seoul drinking water — ibuprofen, phenytoin (Dilantin), carbamazepine, caffeine, the flame retardant TCEP, and the insect repellent DEET. GAC filtration reduced every one of them below practical detection limits, which were 10 ng/L for TCEP and caffeine and 1 ng/L for the rest. RO and nanofiltration did the same in all cases but one: TCEP fell from 284 ng/L to 14 ng/L under RO and 13 ng/L under NF — roughly 95% reduction rather than complete removal.

Why RO rejection isn't one number

"RO removes 99% of pharmaceuticals" is close enough to true to be useful and wrong enough to be worth correcting. Rejection is compound-specific, and the variation is large.

A study of clean and fouled nanofiltration membranes across nine pharmaceuticals and five endocrine disruptors found rejection varied by both compound class and membrane tightness. A tight NF-90 membrane rejected hydrophobic neutral compounds at 95–98%, almost entirely by size exclusion, and ionic compounds at about 99% via electrostatic repulsion. A looser NF-200 membrane rejected hydrophilic neutral compounds at only 35–70%, and ionic compounds at 71–94%. Fouling with organic material degraded the looser membrane's performance further on hydrophilic compounds.

Three mechanisms are operating simultaneously, and which one dominates depends on the molecule:

• Size exclusion — the molecule is physically too large to pass. Dominant for larger, hydrophobic neutral compounds, and the most reliable of the three.
• Electrostatic repulsion — charged membrane surface repels like-charged ions. Highly effective for ionic pharmaceuticals, which is why charged compounds often show the highest rejection figures.
• Hydrophobic adsorption — the compound partitions into the membrane. This can look like high rejection early in operation and then decline as the membrane saturates, which is a real trap in short-duration testing.

The disinfection byproduct question

Chlorination doesn't just fail to remove many pharmaceuticals — it reacts with them. Chlorine and chloramine oxidize organic compounds and produce disinfection byproducts, and the transformation products of pharmaceutical compounds are far less characterized than the parent compounds.

Researchers commenting on the finished-water pharmaceutical data raised precisely this point: the unknown effect of adding DBPs to a mixture of drug residues, and whether that combination could enhance toxicity relative to either alone. The scale comparison is worth holding onto — DBPs are typically found at microgram-per-liter levels, roughly 1,000 times higher than the nanogram-per-liter concentrations reported for pharmaceuticals in finished water. Whatever the pharmaceutical mixture question turns out to be, DBPs are the larger mass by three orders of magnitude, and unlike pharmaceuticals, several DBP classes are already regulated with enforceable MCLs.

This is also why the CCL 6 grouping is coherent rather than arbitrary: disinfection byproducts and pharmaceuticals are listed together as two of the four groups, and they're chemically entangled. For the treatment side of chloramine and DBP control, see the chloramine in water guide, which covers catalytic carbon EBCT requirements and why standard GAC is inadequate for chloramine.

Where this matters commercially

For most commercial facilities, pharmaceutical residues at nanogram-per-liter concentrations are not an operational problem, and treating them as one is a misallocation of capital. There are specific exceptions where the water quality specification is stringent enough that trace organics matter:

• Dialysis — AAMI/ISO water quality standards for hemodialysis are driven by the extraordinary exposure volume. A dialysis patient's blood contacts on the order of hundreds of liters of water weekly across a semipermeable membrane, versus a couple of liters ingested. Dialysis water systems already run RO with redundant treatment for this reason, and trace organics fall within an existing framework rather than requiring a new one.
• Pharmaceutical and biotech manufacturing — USP Purified Water and Water for Injection specifications control total organic carbon tightly. Trace pharmaceutical contamination in feedwater is a cross-contamination concern in a compliance context, not a consumer health question.
• Laboratory water — ASTM Type I water for analytical work, particularly LC-MS applications, where a trace organic in the reagent water becomes a background peak. See the deionized water guide for the purity type breakdown.
• Facilities on wastewater-influenced source water — a surface intake downstream of a treatment plant discharge, or a shallow well near a septic field. This is where source characterization is worth actual money, because the local occurrence data is what determines whether anything needs doing.

For food service, hospitality, and general commercial applications, the honest assessment is that if you're installing RO for scale control, taste, or TDS reduction — which is the usual driver — you're already getting pharmaceutical rejection as a side effect of equipment you were buying anyway. That's a reason to specify the system properly, not a reason to buy a second system.

Testing — and why field instruments can't do it

This deserves stating plainly because the industry is not always clear about it: no field colorimeter, photometer, or portable spectrophotometer can measure pharmaceutical compounds in water. Not the Hach DR300, not the DR900, not the DR1900, not any comparable instrument from any manufacturer.

The reason is fundamental to the method. Colorimetric analysis works by reacting a target analyte with a reagent that produces a color change proportional to concentration, then measuring absorbance at a specific wavelength. It requires a selective reagent chemistry for each analyte and typically has detection limits in the parts-per-billion to parts-per-million range. Pharmaceutical compounds occur at parts per trillion, there is no selective colorimetric reagent for carbamazepine or meprobamate, and dozens of structurally similar compounds coexist in the same sample.

Measuring these compounds requires liquid chromatography with tandem mass spectrometry — chromatographic separation followed by mass-based identification and quantification. That's a laboratory instrument costing well into six figures, operated by trained analysts, with solid-phase extraction sample prep and isotopically labeled internal standards. It is not a field measurement and won't become one.

FAQ

Are pharmaceuticals regulated in US drinking water?

No. As of July 2026 there is no Maximum Contaminant Level for any pharmaceutical compound under the Safe Drinking Water Act. Pharmaceuticals appear as one of four chemical groups on the draft CCL 6 published April 6, 2026, with the final list expected to be signed by November 17, 2026 — but a CCL listing is a queue position, not a regulation, and imposes no requirements on water systems.

Is it true that 80% of US drinking water contains pharmaceuticals?

No. That figure comes from a reconnaissance of targeted streams, not drinking water, and counted detections of any organic wastewater compound — detergents, sterols, flame retardants, caffeine — not just therapeutic drugs. The drinking water source data found detections in about 6% of public-supply well samples and 11% of domestic-supply samples, with a single sample out of roughly 1,100 exceeding a human-health benchmark.

Does my Brita or pitcher filter remove pharmaceuticals?

Only if it carries NSF/ANSI 401 certification, and most don't. Standard pitcher filters are typically NSF 42 — chlorine taste and odor only. Even where carbon is the right removal mechanism, shallow media beds don't provide the empty bed contact time that effective adsorption requires. Check the specific product's certification listing rather than the packaging claims.

Does reverse osmosis remove pharmaceuticals?

Yes, for most compounds, and it's the most reliable single barrier available at point of use. Published research shows RO, nanofiltration, and properly sized GAC each removing more than 99% of many targeted pharmaceuticals, with most compounds reduced below practical detection limits. The exceptions are small, polar, uncharged molecules, where rejection can drop to around 95% or lower. Rejection also varies with membrane tightness and degrades somewhat with fouling.

Does boiling water remove pharmaceuticals?

No — it makes the concentration worse. Pharmaceutical compounds are non-volatile. Boiling removes water as vapor and leaves the compounds behind in a smaller volume. The same is true for nitrates, lead, and most dissolved solids. Boiling addresses microbiological contamination and nothing else.

Should a commercial facility be treating for pharmaceuticals specifically?

In most cases, no. There's no regulatory driver, no MCL to comply with, and detected concentrations in finished water are generally at parts-per-trillion levels below established health benchmarks. The exceptions are dialysis, pharmaceutical and biotech manufacturing, analytical laboratory water, and facilities drawing from demonstrably wastewater-influenced sources. If you're already installing RO for scale, TDS, or taste, pharmaceutical rejection comes with it.

What happens to unused medication that gets flushed?

It reaches a wastewater treatment plant that wasn't designed to remove it, and a fraction passes through to the receiving stream. Conventional secondary treatment removes some compounds through biodegradation and sorption to biosolids, but stable compounds like carbamazepine largely survive. Take-back programs exist specifically to keep pharmaceuticals out of the waste stream at the source, which is by a wide margin the cheapest point of intervention — removing a compound at nanogram-per-liter concentration from millions of gallons is orders of magnitude more expensive than never introducing it.

Sources: US Geological Survey, "Hormones and Pharmaceuticals in Groundwater Used as a Source of Drinking Water Across the United States," Environmental Science & Technology; USGS National Water-Quality Assessment Project; USGS National Reconnaissance of Pharmaceuticals, Hormones, and Other Organic Wastewater Contaminants in Streams; EPA Office of Research and Development, "Determining the Prevalence of Contaminants in Treated and Untreated Drinking Water"; EPA Draft Sixth Contaminant Candidate List, 91 FR (April 6, 2026); EPA Final Fifth Contaminant Candidate List (November 2022); Oklahoma State University Extension, "Pharmaceuticals in Drinking Water"; peer-reviewed studies on nanofiltration and reverse osmosis rejection of pharmaceutically active compounds; NSF/ANSI Standard 401. This guide is informational and does not constitute engineering, regulatory, or medical advice.