New Zealand schools that supply their own drinking water have been drinking water suppliers in law since the Water Services Act 2021, 71 of them reported at least one E. coli detection in 2024, and the assessment contracts covering every self-supplied state school did not commence until 2025-11-01

What is happening?

About 20 percent of New Zealand schools, roughly 440 of them, are not connected to a reticulated town supply. They draw drinking water from rainwater tanks or from groundwater bores on their own grounds, and under the Water Services Act 2021 each of those schools is itself a drinking water supplier, carrying the full set of legal duties that come with that status. Most of them are small and rural, and the group includes kura.

The regulator publishes what that has looked like. In 2024, 71 self-supplied schools reported at least one detection of E. coli, which is an indicator of faecal contamination. 24 of them had repeat detections. Those 71 schools accounted for 53 percent of all E. coli detections notified to the regulator that year.

Through calendar 2025, 45 schools held a do-not-drink notice. Across that year no additional filtration was installed at any of them, and the responsible department held no estimate of what remediation would cost. Reporting from April 2026 puts about 9,000 students at more than 70 schools and early learning services where drinking water breached the standards at least once during that year.

An official information response reported in April 2026 records that the education portfolio had received no advice on the do-not-drink notices issued to schools during 2025, held no cost estimate for remediation, and had exchanged no correspondence on the matter with the local government and health portfolios. The same response describes the matter as operational.

In the twelve months to the June 2026 announcement, 66 schools recorded 85 breaches of drinking water safety, with E. coli, lead, manganese and iron among the contaminants found.

Between 8 and 11 June 2026 the government announced a Drinking Water in Schools Programme funded at NZD 30 million across the 2026 and 2027 financial years. At the announcement 230 of the roughly 440 schools were participating, 64 had begun urgent works, and 20 had a completed upgrade. The stated targets are to assess about half of the self-supplied schools by the end of 2026, to complete upgrades at 98 schools by the middle of 2027, and to have every self-supplied school on safe drinking water infrastructure by 2030.

Whose problem is this?

The duty and the money sit in different places. A school board is the drinking water supplier in law, the property it would have to upgrade belongs to the Ministry of Education, the rules and the notices come from Taumata Arowai, and the network some of these schools might otherwise join belongs to a council. No opened source names a single body that is answerable for the gap between the duty and the plant.

Where does this problem end?

The boundary matters here because the duty is not in dispute and the rules already exist. A school that draws its water from a tank on its own grounds is a drinking water supplier in law, and what is missing is not a rule but the sequence of assessment, funding and installation that turns that rule into a working treatment plant on the ground.

What is the state now, and what should it be?

Now

What it should be

The target state here is written by the bodies responsible and needs no outside standard. The Water Services Act 2021 already sets the floor: water supplied at these schools is to meet the drinking water standards, because the school is a supplier. On top of that floor the government has published dates. About half of the self-supplied schools are to be assessed by the end of 2026. Upgrades at 98 schools are to be complete by the middle of 2027. Every self-supplied school is to be on safe drinking water infrastructure by 2030.

The middle target is not stated consistently across the sources opened here. One opened report of the same announcement carries 98 schools upgraded by mid-2027, and another, describing agreement with the regulator, carries 64 schools at full compliance by the end of June 2027. No opened source says whether those two describe the same set of schools or two different tests, one being a completed physical upgrade and the other a regulatory compliance status. Both are recorded and neither is merged into the other.

How big is it?

About 440 schools, roughly one in five in the country, supply their own drinking water. That is the population the 2030 target covers, and at the June 2026 announcement 20 of them had a completed upgrade.

The measured harm is counted differently in each source and each count covers a different window. 71 schools reported an E. coli detection during calendar 2024, 24 of them more than once. 45 schools held a do-not-drink notice through calendar 2025, with no additional filtration installed at any of them that year. 66 schools recorded 85 breaches in the twelve months to June 2026. About 9,000 students were at more than 70 schools and early learning services with at least one breach during 2025.

These counts cover different periods and no opened source states how far they overlap. The 45 schools under notice in 2025 are not stated to be a subset of the 66 that recorded a breach in the year to June 2026, and neither set is mapped onto the 71 schools of 2024. The figure of about 9,000 students mixes schools with early childhood services, which run on a different funding path from the school property portfolio, and no opened source splits it. A separate 2024 tally reported in the same round covers suppliers serving a population of about 80,000, of which 38 did not issue a public warning after an E. coli positive and 18 of those were small rural schools; that tally covers small suppliers generally rather than schools alone, and its overlap with the school counts above is not established in any opened source.

The number of people is not derivable. No opened source gives the enrolment of the roughly 440 self-supplied schools, so the count of people who drink from an unreticulated school supply has no second term. The one student figure available, about 9,000 for 2025, counts only institutions where a breach was confirmed in that single year and mixes two sectors, so it is neither the exposed population nor a subset that can be scaled up to it. The population file records this as not-derivable with the missing terms named, rather than multiplying 440 schools by an assumed average roll.

Under what conditions does it arise?

1. The supplier is the school. The Water Services Act 2021 puts the drinking water supplier duty on the school itself. A school board can be given a duty, but it cannot appropriate a capital budget, and the property it would have to alter is Crown property administered by the Ministry of Education. The duty and the means to discharge it were separated at the moment the duty was created.

2. The regulator and the funder are both the Crown. Taumata Arowai regulates a set of suppliers that includes state schools, whose buildings and budget belong to a government department. No source opened here explains what enforcement means in that configuration, or what a regulator is expected to do when the party that would have to pay for compliance is another arm of the same government.

3. The remedy is distributed across portfolios, and nothing forces them to meet. Property sits with education, water rules sit with the regulator, and the reticulated networks that some of these schools could join sit with councils under local government. The official information response reported in April 2026 records no correspondence between the education, local government and health portfolios on this matter.

4. A notice manages the risk without repairing the asset. A do-not-drink notice tells people not to drink the water. Nothing in the sources opened here attaches a repair deadline to such a notice, and 45 schools held one through a full calendar year with no additional filtration installed in that year.

5. Small rolls make treatment expensive per student. The schools that are not on a network are mostly the small and rural ones, so each needs its own plant, its own testing regime and its own operator competence, and there is no larger supply to share the fixed cost with. The same characteristic that puts a school outside the network puts it outside the economies that make treatment cheap.

6. Nothing visible happens when the gap persists. Taps keep running, the notice stays on the wall, and the roll does not change. A year in which no filtration is installed at 45 schools produces no event, no publication and no deadline, which is why the state of these supplies became public through a regulator page and an official information response rather than through anything happening at a school.

What has been tried?

Two things were tried in sequence rather than together. The duty arrived in 2021, and the assessment that would tell anyone what the duty costs to discharge began in November 2025. The programme that funds the work was announced seven months after that, with its first published targets falling in 2026 and 2027.

What was found?

Why is it still unsolved?

Coordination failure — the law, the regulator, the rules and the responsible department all exist, and the thing that does not exist is a single point at which the duty, the money and the works meet.

The first part is that the duty was placed on the party least able to discharge it. A school board became a drinking water supplier in 2021 by operation of the Act. What a supplier has to do is well defined, and a board that wants to comply has to obtain capital works on a building it does not own, from a budget it does not control, on a timetable set by a department that answers to a different set of pressures. Nothing about that arrangement is hidden or unusual. It simply means that the entity carrying the legal duty is not the entity that can act on it.

The second part is that the regulated party and the funder are the same Crown. Taumata Arowai can issue a do-not-drink notice, and 45 of those notices sat in place for a full calendar year while no additional filtration went in. No opened source describes what escalation is available when the supplier is a state school and the money would have to come from a government department. A notice is a warning and not a repair, so the number of schools under a do-not-drink notice can stay flat for a year without anything in the plumbing changing.

The third part is that nobody was assembling the picture. The official information response reported in April 2026 records no advice received on the 2025 notices, no cost estimate held, and no correspondence with the two other portfolios whose functions touch the remedy. Assessment contracts to find out what the roughly 440 schools actually need did not commence until November 2025, about a year after the regulator figures for 2024 were public. Until that work reports, the total cost of the 2030 target is a number that no source opened here can state, which means the programme funded at NZD 30 million is not measurable against the size of the task it is meant to complete.

The fourth part is that the gap produces nothing anyone can see. A completed upgrade is invisible from the road, and so is a missing one. The tap runs either way. The only artefacts this problem generates are a line on a regulator page, a notice on a wall and an answer to a written question, and none of those arrives on a schedule that forces a response.

What observation would mean it is solved?

Candidates — (a) every self-supplied school has treatment infrastructure that meets the drinking water standards, with per-school status published (b) the count of schools holding a do-not-drink notice reaches zero and stays there across consecutive years (c) the annual count of self-supplied schools recording a breach falls and keeps falling.

(a) alone counts assets and not water. Installed infrastructure has to be operated, tested and maintained by a school that is also running a school. A completed upgrade recorded in 2027 says nothing about whether the plant is working in 2031, and no opened source describes what ongoing operation, testing and competence the programme funds after installation.

(b) alone can move without any works. A notice can be lifted after a clear retest following a transient event, and it can also be avoided entirely by a supply that is never tested. The count of notices measures the intersection of contamination and detection, so it falls both when water gets safer and when testing gets thinner.

(c) alone is sensitive to how hard anyone looks. These schools accounted for 53 percent of all E. coli detections notified to the regulator in 2024, which is a statistic about detections rather than about contamination. More testing at more schools would push the breach count up while the water improved, and a drop in testing would push it down while nothing improved.

The three have to be read together, and all three would still leave the denominator missing. Until some source publishes enrolment at the self-supplied schools, every observation available here counts institutions and works, and none of them counts the people who drink the water.

What is it connected to?

Fills with researchthe Local Water Done Well reform of council water services and its investment gap, freshwater and nitrate regulation by regional councils as the upstream source of school water contamination, rural school property funding and the small-school cost structure generally, drinking water compliance among small suppliers other than schools, and the same school water supply question in other jurisdictions. Relation type and evidence grade were not established in this round.

What these sources do not say

• What reaching the 2030 target costs. No opened source gives a total remediation estimate for the roughly 440 schools, and the official information response reported in April 2026 records that no departmental estimate was held. Without that figure the NZD 30 million programme cannot be read as sufficient or insufficient by anything in this file.
• How many students attend the self-supplied schools. No opened source gives enrolment for those schools, either per school or as a national total, so the exposed population has no published number anywhere in this file.
• How the school counts relate to one another. The 71 schools of 2024, the 45 schools under notice through 2025, the 66 schools of the year to June 2026 and the more than 70 institutions holding about 9,000 students in 2025 are each published on their own. No opened source states how far these sets overlap or whether any is a subset of another.
• Whether the mid-2027 target is 98 schools or 64. Two opened reports of the same announcement give different figures for the same period, and neither states its basis or acknowledges the other.
• What enforcement means when the supplier is a state school. No opened source describes what steps are available to the regulator beyond issuing a notice, or what happens if a notice remains unremedied across several years.
• The school-level detail behind the 2024 count. The regional distribution of the 71 schools, the reasons for repeat detections, and the rate at which self-supplied schools have filed drinking water safety plans are reported by a source that returned an automated retrieval failure in this round, so none of those figures is carried here.
• The official information response in its own terms. No opened source reproduces the response itself. Its contents reach this file through reporting of it, so the wording of the questions asked and the answers given cannot be checked here.
• What the roughly 440 figure is counted from. Several opened sources give about 440 or about 20 percent, and none of them cites a register or names the count that produced the number.
• Whether private schools face the same position. They hold the same supplier duty under the Act, and no opened source gives figures for them or states how their remediation is funded.
• What the government says about the years before the programme. The opened government release announces the programme and its targets. No opened source carries a departmental account of why no additional filtration went in during 2025 at the schools under notice.

See the evidence

No primary legal document and no regulator report was read in full. The Water Services Act 2021, the Drinking Water Regulation Report 2024 and the official information response itself are described here only through pages and reporting that cite them. Three government pages were opened directly and carry the load for the legal position, the 2024 school figures and the assessment contracts: the departmental school water supply guidance, the regulator guidance page for schools, and the government tender record. Everything about the 2025 notices, the cost estimate and the correspondence reaches this file through one report of an official information response, which is why those rows are graded medium rather than high. Where sources overlap they agree: the figure of about 440 schools or about 20 percent appears in three opened sources, the NZD 30 million programme and its financial years appear in two, and the 2030 target appears in two. Where they disagree the disagreement is left visible rather than resolved — the mid-2027 target is 98 schools upgraded in one opened source and 64 schools at full compliance by the end of June 2027 in another, and no opened source says whether those describe the same set. The separate 2024 supplier tally covering a served population of about 80,000 is recorded as a different population, because no opened source establishes its overlap with the school counts. This is a Path A output, research-based definition, so observation_refs is empty and provenance_mode: press-derived.

People affected

Estimated range Not derivable

The reason and what is missing are listed under “What is missing” below