Water filter shopping is contaminant-driven, not brand-driven. The first question is what you are trying to remove — chlorine taste, lead, PFAS, sediment, bacteria, or fluoride — and the right answer differs for each. This page is structured around that decision: identify your contaminants, match to a filter technology that targets them, and pick the form factor that fits your house and your tolerance for maintenance.
What are you actually trying to remove?
Most US water filter buyers start with the wrong problem. The municipal water in 90% of US homes is safe to drink straight from the tap — the chlorine is doing its job and the inorganic contaminants are below EPA limits. The reasons to filter:
- Aesthetics — chlorine taste, mineral cloudiness, occasional metallic notes. The most common buyer motivation. Solved by activated carbon.
- Lead — from aging service lines (especially in cities with pre-1986 plumbing infrastructure: Chicago, Detroit, Newark, Pittsburgh, parts of NYC). Solved by NSF/ANSI 53-certified lead-reduction filters.
- PFAS / forever chemicals — emerging concern, particularly in areas downstream of industrial sites, firefighting training facilities, and military bases. Solved by certain activated carbon filters (must be NSF/ANSI 53 P473 certified) or reverse osmosis.
- Sediment — rust, silt, grit from older pipes or well water. Solved by 5–50 micron sediment cartridges, usually installed at the point of entry.
- Microbes — bacteria, cysts (Giardia, Cryptosporidium). Real concern on private wells and unreliable municipal supplies. Solved by UV sterilization or sub-micron filtration.
- Hardness — calcium, magnesium. Not what a filter does. You need a softener for this.
- Fluoride — added by municipal utilities for dental health. Removable only by reverse osmosis or specialized fluoride-removal media (not standard carbon).
Get your annual Consumer Confidence Report (CCR) from your water utility — every US utility is required to publish one. Read what is actually in your water before choosing a filter. The most expensive mistake in this category is buying a filter optimized for a problem you do not have.
Five distinct form factors, each with its own use case:
- Pitcher filters (Brita, PUR, ZeroWater, LifeStraw Home): $25–$60 upfront, $5–$15/month in cartridges. Activated carbon (most brands) or ion exchange (ZeroWater). Right answer for renters, dorms, or trial-mode users. Filters one pitcher (roughly 8–10 cups) before exhaustion — heavy household use will exhaust a Brita cartridge in 2–3 weeks.
- Faucet-mount filters (PUR, Brita on Tap, Culligan): $25–$45 upfront, $20–$30 per quarterly cartridge. Activated carbon, often NSF 53 certified for lead. Convenient until they leak from the diverter mechanism — typical life 1–2 years before the unit itself needs replacement.
- Under-sink point-of-use (Aquasana, APEC, iSpring): $150–$400 installed. Multi-stage carbon, often combined with sediment pre-filter. The right answer for a primary drinking-water tap when you do not need to filter the whole house.
- Whole-house point-of-entry (Aquasana Rhino, SpringWell CF, US Water Systems): $700–$2,500 installed. Treats all water entering the house: showers, dishwasher, laundry, drinking. Right answer if your concern is chlorine smell in showers, or if you want PFAS reduction at every tap.
- Reverse osmosis under-sink (APEC ROES-50, Waterdrop G3, iSpring RCC7): see the dedicated RO category. Removes more contaminants than activated carbon but produces ~3 gallons of waste per gallon of treated water and removes beneficial minerals along with the bad.
The choice is not which is best in absolute terms — it is which matches your contaminant target and your tolerance for installation complexity.
The certification system — what NSF actually means
The NSF/ANSI certification standards are the only useful filter certifications. Marketing language is unreliable; certification language is verifiable.
- NSF/ANSI 42: aesthetic improvement — chlorine, taste, odor. Almost every filter claims this. Means little for safety.
- NSF/ANSI 53: health-effect contaminants — lead, mercury, asbestos, VOCs, cysts. The standard that matters for safety claims. Specific contaminants are listed individually within the standard — a filter "53-certified for lead" may not be certified for mercury.
- NSF/ANSI 58: reverse osmosis system performance.
- NSF/ANSI 401: emerging contaminants — pharmaceuticals, herbicides, some industrial compounds. Newer standard, fewer certified products.
- NSF/ANSI P473: PFOA and PFOS (two specific PFAS compounds). Increasingly important.
- NSF/ANSI 372: lead-free materials. Required for any filter used in drinking water in the US.
"NSF tested to" without "certified" is marketing weasel-wording. The certification is the audit. Brands that claim "tested" without certification are essentially saying "we ran a test we cannot show you."
Micron rating — what the numbers mean
Sediment filters and ceramic filters are rated in microns — the smallest particle size the filter blocks. Roughly:
- 50 micron: visible sediment, sand, larger rust particles.
- 20 micron: most sediment, smaller rust, organic debris.
- 5 micron: fine sediment, most cyst-size organisms.
- 1 micron: bacteria, most fine sediment.
- 0.5 micron: cysts (Giardia, Cryptosporidium).
- 0.2 micron: most bacteria (E. coli, Salmonella).
- 0.01 micron (ultrafiltration): viruses.
Each step down increases flow restriction. A 0.2-micron whole-house filter cannot deliver the same GPM as a 50-micron sediment filter at the same pressure — physics is unavoidable. Match micron rating to actual threat, not to the most-impressive number.
How often filters actually need changing
Manufacturer-recommended filter change intervals are usually conservative; real-world filter life varies wildly with water quality.
| Filter type | Manufacturer interval | Real-world interval (good water) | Heavy contaminant interval |
|---|
| Pitcher (Brita) | 2 months / 40 gallons | 6–8 weeks | 3–4 weeks |
| Faucet-mount | 3 months | 3 months | 6 weeks |
| Under-sink carbon | 6 months | 9–12 months | 4–6 months |
| Whole-house sediment | 6–9 months | 9–12 months | 3–4 months |
| Whole-house carbon | 12 months | 12 months | 6 months |
| RO membrane | 2–3 years | 3–5 years | 18 months |
The reliable indicator of exhaustion is taste or flow degradation, not the calendar. If the filter still tastes neutral and flow has not slowed, it has remaining life regardless of date. If taste has returned to "chlorine" or flow has noticeably reduced, the filter is exhausted regardless of date.
Brand tiers
The major brand groups, in tiers of capability:
Tier 1 — premium / certified for health-effect contaminants: Aquasana, Multipure, Berkey (gravity-fed), AquaTru (countertop RO), APEC. These hold NSF 53 and 401 certifications, publish their certified contaminant lists, and have established service histories.
Tier 2 — mainstream / NSF 42 certified, sometimes 53 for chlorine and lead: Brita, PUR, ZeroWater, Culligan, Pelican, GE. Reliable for chlorine and taste; check the spec sheet for any contaminant-specific claims.
Tier 3 — generic / unverified: the Amazon long tail of $20–$60 filters from brands you have never heard of. Some are legitimate (often re-badged from a Tier 2 OEM); many are not. Hard to tell without third-party verification. If health claims matter, do not buy from this tier.
10-year cost reality
A common surprise on filter ownership is the cumulative cartridge cost over a decade — usually 2–4× the original filter purchase price.
- Brita pitcher: $40 device + $120/year cartridges = $1,240 over 10 years.
- PUR faucet-mount: $35 device + $100/year cartridges + 1 device replacement = $1,070 over 10 years.
- Under-sink three-stage carbon: $300 device + $90/year cartridges = $1,200 over 10 years.
- Whole-house Aquasana Rhino: $1,200 device + $120/year cartridges + a $250 mid-life replacement = $2,650 over 10 years.
- RO under-sink: $300 device + $80/year filter changes + $80 mid-life membrane = $1,180 over 10 years.
The form factor with the lowest per-decade cost surprises a lot of buyers: well-chosen under-sink carbon and entry-level RO both come in slightly under the pitcher path. The pitcher path's per-cartridge cost adds up faster than people anticipate.
Five filter-shopping mistakes
- Buying a filter for contaminants you do not actually have. Read your CCR. Match the filter to the report.
- Confusing softeners and filters. A filter does not remove hardness. A softener does not remove lead. They solve different problems.
- Treating NSF 42 as equivalent to NSF 53. Aesthetic and health-effect certifications are distinct standards. Marketing copy often blurs them.
- Skipping the sediment pre-filter on whole-house systems. A $40 sediment cartridge upstream extends your $300 carbon filter's life by 60%. Always pre-filter sediment in whole-house installs.
- Choosing whole-house for a problem that is only at the kitchen tap. If your only concern is drinking water taste, a $200 under-sink solves it. A whole-house carbon system costs $1,500 to solve the same problem. Match scope to need.