River Restoration: The Missing Link Between Spending and Results

Ask what England spends on restoring its rivers, and what it gets for the money, and the published record cannot answer. The answer is not hidden. It was never assembled. There is no series of what was actually spent on restoration, no audit of that spending, and no English project in the record we read that reports both a cost and a measured ecological result.

What can be said is smaller, and more useful. The two sources that label money as river restoration, the Environment Agency’s restoration funds and the water companies’ restoration lines, come to roughly £50m a year in England between them. Fines and farm schemes add more that nobody counts as a river figure. Restoration is under 1% of the water industry’s environment programme. And the billions that circulate whenever rivers are debated, by their own sources’ descriptions, buy something else.

The billions buy something else

Start with water bills. Ofwat’s price control for 2025 to 2030 is a £104bn package for England and Wales, which the government describes as money “to upgrade crumbling pipes and build new sewage treatment works” (Defra, July 2025). Within it sits the Water Industry National Environment Programme (WINEP), the work water companies in England are required to do to meet their environmental obligations, worth £22.1bn over the five years.

Ofwat’s own line items show what that programme buys. Across England and Wales, where Wales runs an equivalent programme, the allowance comes to £23,891.9m in 2022-23 prices. Storm overflows take £11,940.3m of it, half the total. Nutrient removal takes another £5,966.8m.

Six lines are recognisably restoration: fish passes (£50.0m), river connectivity (£12.5m), habitat restoration (£97.6m), invasive species (£40.9m), eel screens (£28.5m) and fish screens (£1.7m). Together they come to £231.2m, or 0.97% of the programme. Leave out the two Welsh companies and the figure falls to £133.7m, mostly because £86.0m of the habitat restoration line belongs to Dŵr Cymru. Add the broader biodiversity lines and restoration-type spending reaches 2.2%. All of these are allowances: what companies may recover from bills, not what they have spent.

The other public billions are flood defence. England’s flood and coastal erosion programme spent £1,402.0m in 2024/25 alone. Some of it touches rivers. It is bought, and reported, as flood protection.

The figure most likely to be mistaken for restoration money is the government’s statement that water companies “will invest £2 billion over the next 5 years to deliver more than 1,000 targeted actions for chalk stream restoration” (written answer, July 2025). That is 8.65 times the entire restoration allowance for England and Wales in Ofwat’s line items. Most of it must therefore be other programme work described as chalk-stream restoration. What that work is has not been itemised in anything we read.

The same £22.1bn reaches debates about restoration, though not as restoration money. In a January 2026 Westminster Hall debate titled “River Habitats: Protection and Restoration”, an MP welcomed “the Environment Agency securing a record £22.1 billion in investment over the next five years” in the same sentence as the commitment to halve sewage pollution and storm overflow spills by 2030, then turned to Wessex Water’s work at two sewage treatment plants in the constituency (Hansard). That is a fair description of what the money buys. Storm overflows are half of the England and Wales programme it sits in, and restoration lines are under 1%.

What restoration money looks like

On the public side, one list exists. A written answer of 5 March 2026 sets out what the Environment Agency was “allocated several funds in 2025/26 that directly contribute to the restoration of river and coastal habitats”:

£10.9 million, Water Environment Improvement Fund
£3.3 million, Habitat Compensation and Restoration Programme
£8.6m, Environmental Statutory Allowance

That is £22.8m in total. The same answer names £12.6m more for the Water and Abandoned Metal Mines programme, shared with the Mining Remediation Authority, but sets it apart from the funds that “directly contribute to the restoration of river and coastal habitats”, so it is left out here too. Three caveats travel with the list. It is an allocation, not money spent. It includes coastal habitat. And two of its three lines are paid from the flood budget. Set against that year’s flood budget of £1,508.4 million, the whole list is 1.51%.

Add the public list to a fifth of the bill-funded restoration lines excluding the two Welsh companies (£26.7m a year from £133.7m over five years) and the result is roughly £50m a year in England. That sum adds an allocation to an allowance, across two price bases, so it is an order of magnitude rather than a total. It also leaves money out. Water company fines now fund local projects (“Over £10 million” to 51 projects in October 2025, set out below), but those projects mix river and floodplain restoration with studies, surveys and septic-tank engagement. Farm schemes such as Landscape Recovery can include river work, and we found no river share for them; nor did we find a figure for river habitat work funded through Natural England. None of these publishes an annual river-restoration total. The £50m is as precise as the published record allows, and it counts what those two sources label as restoration, not everything spent on or near rivers.

No spending series

To check what was actually spent, rather than allocated or allowed, there is almost nothing to look at. No annual series of restoration outturn exists in any source we read, public or bill-funded. The Environment Agency’s own WINEP datasets have no cost column at all: 83 columns in the current dataset, 96 in the previous one, none of them money.

The only line-level outturn we found covers the first two years of the previous price control. In 2020 to 2022, across England and Wales, four restoration-type lines spent £75.0m against £65.8m allowed (2021-22 prices). That looks like delivery ahead of plan, until the lines are separated: £64.0m of the £75.0m was eel measures at water intakes, which protect fish at abstraction points rather than restore habitat. Invasive species work spent £3.2m against £23.2m allowed for the same two years.

Water company fines show the same gap. The Water Restoration Fund was given up to £11 million in fines. In April 2025 the National Audit Office reported that “The funds remain unused” (NAO). By October 2025, “Over £10 million” had been channelled into 51 projects. In July 2025 Defra said “over £100 million of water fines are being spent on local clean-up projects” (Defra); in January 2026 the figure was £29 million, which a later answer said “constitutes all fines received between October 2023 and October 2025, excluding any enforcement undertakings agreed by regulators” (written answer). The figures may count different things. No source reconciles them, and none says how much has been spent. When an MP asked whether the government would “publicly track i) where and ii) how much fine monies are spent on river restoration”, the answer named channels and did not commit to tracking.

What gets counted instead

Without spend or outcome data, the system counts outputs. The Environment Agency reports that 10,945 km were enhanced between January 2016 and March 2022. It adopted kilometres because classification “results have not always reflected the work being undertaken to enhance the water environment”. That is a fair point about lag. It also means the headline measure of restoration is a measure of activity.

Kilometres cannot rank value. In the Environment Agency’s 2024-25 fisheries report, Water Environment Improvement Fund projects drew £15,526,325 of Agency money and £21,499,240 of match funding. Divide that by the 617km of rivers the report says were enhanced and it comes to about £60,009 per km, though the same money also “protected 471km of rivers”, built 10 fish and eel pass improvements and removed 25 weirs and barriers. In the same report, the Wild Trout Trust line works out at about £3,639 per km “directly enhanced”, co-funding included, and that line also paid for 87 advisory visits and 17 practical workshops. Both figures load every pound onto kilometres. A sixteen-fold gap inside one report does not show that one channel is sixteen times better value. It shows that “enhanced”, which the report never defines, cannot carry a comparison. The Wild Trout Trust is also a recipient: the Agency gave it £90,000 for advisory work and £45,000 for project costs that year.

Delivery is measured the same way. For 2020 to 2025, 97.0% of water and sewerage company schemes met their deadlines. The Environment Agency conducted site inspections on 1% of the 8,780 actions companies said they had completed.

The Agency’s clearest value claim rests on a single example. Its October 2025 document on physical modifications says of weirs: “Their removal is one of the most cost-effective ways to restore ecological function” (Environment Agency). The evidence beneath that line is one named project, Bowston Weir on the River Kent, with no figures. The same document says “more robust data is needed to demonstrate outcomes”, and lists “demonstrating value for money” among the tools still to be developed.

Results without a price

Measured ecological results from restoration do exist, on English rivers and in studies by UK researchers. They are few, and none of them comes with a cost.

Gravel on two Hampshire chalk streams. On the Test and the Itchen, invertebrate taxon richness rose one to two years after restoration, measured against control sites. At two restored sites on the Test, invertebrate abundance and diversity increased over 8 to 9 years of monitoring, with several measures lower in 2016.
Barriers on a River Wear tributary. After barrier removal and fish passes, brown trout density rose from 20.9 to 33.8 per 100 m², and bullhead from 4.6 to 32.6. Salmon stayed absent throughout, and stone loach declined where barriers were removed.
Weak and mixed results, also measured. Two lowland UK schemes showed invertebrate recovery that was “limited and inconsistent”. In a UK team’s study of three channel-restoration projects, average community composition in restored reaches differed only “marginally” from control reaches, though changes in how varied those communities were (their “multivariate dispersion”) “were more robust and indicated greater ecological heterogeneity within restored reaches”.

None of these studies reports what the work cost. In the English projects we found that do report a cost, every outcome is a claim in a press release. The £2.5 million project to remove Dovecliff weir on the Dove, announced by the Environment Agency, opened “over 550 kilometres of watercourse” to fish, and the outcome offered is “We’re already seeing the benefits upstream of the site with the formation of gravel bars and improved habitat for invertebrates, spawning fish and other wildlife”, with no data attached.

That is the missing link at its smallest scale. The projects with a measured result carry no price, and the projects with a price carry no measured result.

Monitoring itself is rare. The River Restoration Centre’s inventory holds over 5,000 UK projects. Most of its entries record a cost, with location, objectives, techniques and dates; only “a small amount of projects also have additional details for example on monitoring or specific costings”. Researchers funded partly by the Environment Agency write that for small projects “it is not economically viable to include extensive monitoring”.

Why the national figure cannot settle it

The obvious test is the national classification. In the 2019 classification, 16% of rivers and canals in England were at good or high ecological status. The classification published on 6 August 2026 gives 594 of 3,905 river and canal water bodies at good or better, 15.2%, counted from the Environment Agency’s own data layer.

The Agency says the two rounds are “not directly comparable”. Historic expert judgements that could not be verified have largely been removed, so “some apparent changes simply reflect differences in available evidence”; and the assessment of heavily modified water bodies is now automated, where “gaps in supporting data have temporarily increased the number of water bodies reported as moderate rather than good ecological potential”. Its classifications are “a periodic status assessment, not designed to track change over time”. So the step from 16% to 15.2% cannot be read as a change in the rivers, and nor can the movement of individual water bodies between the two rounds.

Restoration cannot be read off that figure in either direction. National status reflects every pressure on a water body, from sewage to farm runoff, and restoration money is about 1% of the environment programme. No source attributes any change in status to restoration, and a national figure whose rounds cannot be compared would test nothing about it even if one did. “We spent billions and got nothing” is wrong twice: the billions did not go on restoration, and the national figure could not show what restoration did. (Our earlier analysis of why 16% of English rivers meet good status covers the pressures in more detail.)

No audit

None of the National Audit Office reports we read audits restoration spending, and its most recent report on environmental regulation states its limits plainly: “We did not look at EA or NE’s non-regulatory responsibilities” (NAO, January 2026).

What the NAO has found, it found about the system around restoration:

In 2026, “Defra and the regulators do not systematically evaluate the impact of environmental regulation”, and their performance metrics are “primarily based on detailed inputs rather than broader environmental outcomes” (NAO).
In 2025, “Defra did not assess the cost or deliverability of its ambitions, except for storm overflows”, and over two price controls Ofwat and the Environment Agency “have had limited oversight of whether companies carry out the work as expected” (NAO).
In 2023, Defra recognised that its data “does not establish links between observed trends and drivers of change, such as policy or regulatory interventions” (NAO).

The only NAO cost-against-impact verdict in the water environment that we found is from 2010: “The annual expenditure of £8 million has, to date, had little impact” (NAO). It concerned diffuse pollution work, mostly farming, not restoration.

The case that the money is being weighed

The strongest objection to all of this is that costs are weighed against benefits in England, just not in the form this article asks for.

There is a published cost test, and it bites. Under river basin planning, “Disproportionate expense is a political judgement informed by economic information”, and measures were cut back until benefits exceeded costs. In the 2015 plans this cut costs by £11bn while reducing benefits by only £1bn, at 2021 prices (Environment Agency). The 2022 plans set £50.6 billion of costs, in present value over 37 years, against £67bn of monetised benefits, a benefit-cost ratio of 1.33. The limits are stated in the same document: the benefits are modelled, the appraisal covers every sector together, and it “is not a cost benefit analysis considering alternative options”.

Rivers have improved on the measures with long records. At principal-river monitoring sites in England, mean ammonia has fallen 85%, BOD 46% and orthophosphate 83% since 1990 (Environment Agency). The source sets those falls beside reduced loads from sewage works and farms; it does not credit physical restoration. River invertebrates in England and Wales show “an aggregate increase” in richness and pollution-sensitive groups over the past 30 years (Johnson et al.). A separate study, of invertebrate records from 22 European countries, found that the increases it measured “primarily occurred before the 2010s, and have since plateaued”.

An evaluation is under way. In 2022 the WINEP methodology promised that “An evaluation framework will be developed to understand the actual cost-effectiveness of WINEP actions”. The Environment Agency’s contractor now describes “delivering the multi-year evaluation of the WINEP 2025-2030”. We found no published results. Not evaluated through 2025, under way since, is the fair description.

Lag and cost are real constraints. The Agency says, fairly, that “There is a natural lag between a measure being delivered, an ecological response occurring, and that improvement being reflected in a classification”, and researchers working with it judge extensive monitoring of small projects not economically viable. The Office for Environmental Protection’s complaint is chiefly “insufficient investment” and plans “too generic”, not waste.

None of this closes the gap. A modelled benefit-cost ratio across every sector is not a measured return on restoration. Improved chemistry sits, in its own source, beside falling loads from sewage works and farms, not beside restoration. And the absence of a measured link is not evidence that restoration fails. It is the absence of evidence either way, which is the point.

What would close the gap

The pieces exist separately. Ofwat publishes line items. The Environment Agency classifies every water body. A few research teams measure restoration sites before and after. Nothing published yet puts them together.

The Environment Agency’s WINEP evaluation, when published, is the nearest thing to an answer on the bill-funded side. The government’s 2026 water white paper promises to “evaluate the effectiveness of interventions from an ecological and public health perspective”. The Office for Environmental Protection has asked Defra for “a coherent and nested monitoring and evaluation framework for the state of the water environment and progress on measures to improve it”, and government has published a response.

Until one of those produces numbers, the honest answer to what England gets for its river restoration money is that it has not been measured against what was spent. That is a smaller claim than “billions wasted”. It is also the only one the evidence supports.