A peer-reviewed study tested water before and after RO treatment in 216 households and found the microbial reduction was statistically insignificant. The membrane isn't the problem. Maintenance is.
Reverse osmosis systems have become the household water treatment technology of choice for hundreds of millions of people. They are purchased for a specific promise: a membrane so fine it strips water of dissolved salts, heavy metals, and biological contaminants, delivering water at the tap that is safer than what came from the source. A new peer-reviewed study from 216 households in Chennai, India challenges that promise with some of the most uncomfortable empirical data the household water treatment field has yet produced — and its implications extend well beyond India.
The study, published in May 2026 in the Journal of Exposure Science and Environmental Epidemiology by a multi-disciplinary team from Tel Aviv University, IIT Madras, and local partner institutions, is one of the first empirical evaluations of household-level RO filtration effectiveness conducted at this scale under real-world conditions. Its headline finding: after passing through an RO system, nearly one in three water samples still contained E. coli.
The research was conducted under the People's Water Data (PWD) initiative, collecting 262 water samples from 216 households across 23 locations in Chennai. Crucially, samples were collected in paired fashion: one from pre-RO source water and one from post-RO drinking water at the point of consumption. That direct before-and-after comparison within the same household eliminates most of the confounding that weakens observational studies.
The findings are not a simple indictment of RO technology. The study's authors are explicit about this. For the parameters RO membranes are physically designed to address — TDS, conductivity, turbidity, water hardness, and alkalinity — the improvements were dramatic and statistically unambiguous, with p-values near zero. The systems are doing exactly what a functioning RO membrane does.
| Metric | Pre-RO (untreated) | Post-RO (treated) | Statistical significance |
|---|---|---|---|
| E. coli presence | 71% of samples | 31% of samples | p = 0.52 — NOT significant |
| Total coliform presence | 88% of samples | 74% of samples | p = 0.73 — NOT significant |
| Water hardness (mean) | 0.69 ppm | 0.08 ppm | p ≈ 0 — highly significant |
| Alkalinity (mean) | 0.825 ppm | 0.225 ppm | p ≈ 0 — highly significant |
| TDS, conductivity, turbidity | Higher | Significantly reduced | p ≈ 0 — highly significant |
The problem is what those p-values mean together. A p-value of 0.52 on E. coli reduction means the study cannot statistically distinguish the post-RO contamination rate from random variation. The RO systems in these households reliably improved the physical and chemical character of the water while failing to produce a statistically demonstrable reduction in the biological contamination that poses the most direct threat to human health.
A note on the US context: Chennai's source water contamination rates — 71% E. coli positivity before treatment — are not comparable to US municipal tap water. This study's direct health implications are most acute for areas with compromised source water. However, the maintenance failure modes documented here apply universally. As the study's authors explicitly state: "no single household treatment technology guarantees microbiological safety without proper maintenance and safe handling practices — a statement that is as relevant to an Iowa farm family with an RO system under the kitchen sink as it is to a Chennai household."
The study identifies several mechanisms through which post-RO contamination occurs — and almost none of them are about the membrane itself.
Biofilm formation is the first and most technically significant pathway. When microorganisms colonize the interior surfaces of an RO system — tubing, fittings, storage tanks, dispensing taps — they form extracellular polymeric matrices that shield bacterial colonies from cleaning, resist hydraulic flushing, and shed viable pathogens directly into treated water. Research cited by the authors found that after 25 days of initial biofilm formation on installation materials, the biofilm fully covers the surface and is not removed by running water or standard cleaning. It requires physical disinfection or component replacement.
Post-treatment handling practices are the second major pathway. Wide-mouth collection containers, transferring water between vessels, and dipping hand-held utensils rather than dispensing directly from a tap are all documented recontamination pathways. A technically clean water output and a bacteriologically safe cup at the point of consumption are not the same thing.
System age and maintenance interval are the variables with the most direct and measurable impact on outcomes — which brings us to the study's most practically actionable data.
Round 2 of the study, conducted with 226 households approximately one year after Round 1, documented RO system age, filter replacement history, and maintenance intervals. The findings are difficult to interpret as anything other than direct evidence that maintenance drives microbial outcomes.
| System age | E. coli positivity (post-RO) |
|---|---|
| Under 1 year | 6.7% |
| 1–3 years | 29.2% |
| 4–5 years | 27.8% |
| More than 5 years | 66.7% (95% CI: 40%–93%) |
| Last serviced | E. coli positivity (post-RO) |
|---|---|
| Within the past 3 months | 8.3% |
| Within the past year | 69.2% |
| Never maintained | 83.3% |
That last number bears sitting with. An RO system that has never been maintained shows 83.3% E. coli positivity in treated water — statistically indistinguishable from the 71% pre-RO contamination rate in untreated source water. An unmaintained RO system is, by this evidence, essentially non-functional as a microbial barrier. It continues to remove dissolved ions and reduce turbidity — improving water's appearance, taste, and clarity, reinforcing the user's perception that the water is safe — while failing at the biological protection that justified installing it in the first place.
"No single household treatment technology guarantees microbiological safety without proper maintenance and safe handling practices."
— Kagan et al., Journal of Exposure Science and Environmental Epidemiology, May 2026A finding that runs through this study deserves particular attention: the perception of water quality is essentially uncorrelated with its actual biological safety. Among 66 RO-system observations, all 66 produced water described as clear, 61 as odorless, and 46 as tasteless — exactly the sensory characteristics that signal safety to consumers. These characteristics are completely decoupled from whether the water actually contains E. coli.
This creates a false confidence effect in which the presence of an RO system reduces the likelihood of adopting complementary protective behaviors — behaviors that would otherwise serve as a backup safety net. A household using a poorly maintained RO system may be more likely to believe their water is safe than a household using no treatment at all, because RO output looks, smells, and tastes distinctly better regardless of its microbial status.
For US RO owners, this matters in a different way than for Chennai households. The contamination risk from source water is lower in most US contexts — but the false confidence effect still applies. A TDS reading of 9 ppm tells you the membrane is working and removing dissolved solids. It tells you nothing about whether the internal tubing or dispensing tap has developed a biofilm colony.
The study's conclusions are primarily behavioral and educational rather than technical. The authors call for structured maintenance schedules, technician training, standardized servicing practices, and public health communication on safe water handling. The interventions most likely to close the gap between what RO systems promise and what they deliver are downstream of the membrane.
For US homeowners with under-sink or countertop RO systems, the practical implications map directly onto what manufacturers already recommend but few users consistently follow:
Replace filters on schedule — the sediment and carbon pre-filters protect the membrane from fouling, and a fouled pre-filter allows bacteria to reach and colonize downstream components
Most systems: sediment and carbon pre-filters every 6-12 months, RO membrane every 2-3 years, post-carbon polishing filter annually. The study's data suggests maintenance interval is the single strongest predictor of microbial outcome.
Monitor rejection rate with a TDS meter — a new membrane achieves 95-98% TDS rejection; below 85% indicates degradation
TDS monitoring does not detect bacterial contamination, but membrane degradation that allows dissolved solids through also allows bacteria through. Rejection rate is an imperfect but practical proxy for membrane integrity.
Sanitize the system when replacing filters — most manufacturers include a sanitization step in replacement procedures that addresses biofilm risk
Food-safe sanitizers approved for RO systems can reduce biofilm load on interior components. This step is skipped in most DIY filter replacements and is one of the most consistent gaps between manufacturer recommendations and actual user practice.
Dispense directly from the tap — do not dip cups or utensils into the storage tank or drip tray
The post-treatment recontamination pathway documented in the study applies anywhere water is handled after leaving the RO membrane. Direct dispensing into clean vessels at the tap is the simplest behavioral control.
Find your system's exact replacement schedule. Maintenance intervals vary by system, water quality, and usage volume. The generic 6/12/24 month schedule is a starting point, not a guarantee.
→ RO filter replacement schedule by brand and system → How to monitor your RO membrane with a TDS meter
The Chennai study is among the first to put empirical numbers on a problem that has been theorized but rarely measured at household scale: that household water treatment — including the most technologically sophisticated options available — does not guarantee safe water without the behavioral infrastructure to support it.
UV disinfection systems are sensitive to lamp aging and power reliability. Chlorination depends on correct dosing and contact time. Boiling depends on consistent practice and recontamination-free storage. Every household water treatment method that has been tested under real conditions rather than controlled laboratory settings shows the same pattern: performance degrades with inadequate maintenance, and the degradation is not visible to the user because the treated water continues to look and taste acceptable.
An RO system five or more years old with no maintenance history is, by the available evidence, providing little or no microbial protection while continuing to remove dissolved solids and providing clear, odorless, good-tasting water that reinforces the belief that it is working. That gap — between apparent performance and actual performance — is what this study makes measurable.
The membrane works on chemistry, not biology. RO systems produced dramatic, statistically significant improvements in TDS, hardness, alkalinity, and turbidity — but not in E. coli or total coliform. These are different failure modes with different solutions.
Maintenance interval is the primary predictor of microbial outcomes. Systems serviced within 3 months showed 8.3% E. coli positivity. Systems never maintained showed 83.3% — equivalent to untreated source water. This is the most actionable finding in the study.
Biofilm in tubing and components — not membrane failure — is the dominant contamination pathway. The membrane can reject 97% of dissolved solids while internal components harbor bacterial biofilm colonies shedding pathogens into treated water at the tap.
A TDS meter reading of 9 ppm does not mean the water is bacteriologically safe. TDS measures dissolved solids, not biological contamination. Monitoring rejection rate is useful for tracking membrane integrity but does not detect biofilm-derived bacterial contamination.
The solution is primarily behavioral. Structured maintenance schedules, proper system sanitization at filter changes, and safe handling practices are the interventions that close the gap — not hardware upgrades alone.
RO Filter Lab covers reverse osmosis filtration and drinking water quality. This article is based on a peer-reviewed study published in May 2026. The full study is linked below. We have no affiliation with the researchers or institutions involved.
1. Primary study — Kagan, S., Hamilton, B., Lichtman, T., et al. "Evaluating Household Reverse Osmosis Systems for Microbial Safety: A Case Study from Chennai, India." Journal of Exposure Science & Environmental Epidemiology. Published online May 13, 2026.
doi.org/10.1038/s41370-026-00911-52. People's Water Data Initiative — Tel Aviv University / IIT Madras. Community water quality monitoring program.
peopleswaterdata.org3. WHO/UNICEF Joint Monitoring Programme — Global data on household drinking water treatment methods.
who.int