Fly-tipping in England

What the figures count

Apart from the estimate of fly-tipping on farms, the figures start from fly-tipping councils recorded and reported to Defra, mostly on public land. None of them counts fly-tipping on private land other than farms, such as business premises; cases the Environment Agency handles, which are large, organised or hazardous; or reports made to someone that never reached a council’s records, which would need councils’ own records to estimate. Fly-tipping on public land that nobody reported and no council crew found is shown only as a sensitivity (see Public land nobody reports).

How councils record

Each council records fly-tipping in its own way, so a council that records more may have more fly-tipping, record more thoroughly, or both, and the data cannot tell these apart. Every comparison between councils, on the map and in Figures 2 to 4 and, for the councils of the listed wards, Figure 5, mixes what is dumped with how it is recorded: the council views are not scores, and differences within a peer group are differences, not explanations.

Some councils’ records have a break, a boundary change, a change in how incidents are recorded or a data-quality note from Defra. Their figures are shown like any other council’s; when you select one, its note appears in the readout under “The detail: how precise this is”. Each report says where its place stands among all places of its kind in England by recorded rate alone, unadjusted; only the comparisons, which allow for conditions and recording, say whether a council records more than expected. Figure 5 lists wards, with their council, only for councils whose records passed the checks.

Neighbourhood figures

Councils report how much fly-tipping they record, not where. Each neighbourhood figure is the model’s estimate of that neighbourhood’s share of its council’s total, learnt from FixMyStreet reports, local conditions and, in six councils, the council’s own records of where fly-tipping was found; the council’s total is what it recorded, and its neighbourhoods add up to it. So neighbourhood figures, and any ranking of neighbourhoods, such as Figure 2’s, are estimates. Each council’s expected level, on Councils compared with similar councils (what its conditions predict), is a model estimate too.

Testing the model

Six councils’ own records of where fly-tipping was found feed the split within councils: York’s by neighbourhood; Leeds’s, Barnsley’s and Calderdale’s by ward; and Bradford’s and Bassetlaw’s by street or neighbourhood, for earlier years (see Sources). York is the only council checked neighbourhood by neighbourhood, and Leeds was checked by ward. The other four councils’ records are used to place neighbourhoods but have not yet been used as held-out checks, so how the split does elsewhere is not known. Earlier versions were scored on the same York and Leeds records, so the test shows that this version does no worse than the one before it, not how it does on records no version had seen. And the model learns mainly from FixMyStreet reports, a thin record whose pattern partly reflects who reports, although it allows for each neighbourhood’s reporting habit. How the model works, step 3, gives the test and its results.

Large rural neighbourhoods

Large rural neighbourhoods fill most of the map, and their placement is the least certain: checked against York’s own records, the model tended to put too much fly-tipping in large neighbourhoods, more so the larger they are, so some sit in a higher band on the map, or a higher group in Figure 2, than they should. Treat their colours as approximate. For neighbourhoods over 10 km², the readout gives a likely range, which is wider than an 80% interval would be: in the check against York’s records it held for 93% to 97% of neighbourhood-years.

Farm estimate assumptions

Fly-tipping on farms rarely reaches councils’ records, because the landowner usually clears it, and nothing else records it, so it is estimated: the number of commercial farm holdings in each council (Defra’s June Survey, June 2025), shared among its neighbourhoods by their registered land parcels, times the share of holdings hit by fly-tipping in a year, times the number of incidents on each holding hit.

The share is 0.54 (range 0.27 to 0.70), the National Farmers’ Union’s own small-scale figure: 54% of the 620 members who answered its June 2024 survey had had small-scale fly-tipping, and 30% large-scale. The survey was self-selected, which tends to over-represent those affected, so the low value halves it; its results give no period for the figures, so whether they cover a year is not confirmed, and no sampling method. The high value is just short of the Country Land and Business Association’s almost three quarters of responding farmers affected each year, for which no sample size is given. The number of incidents a year on each holding hit is 2 (range 1 to 4). No source gives an average: the Country Land and Business Association reports some farms targeted several times a month, which is the hardest-hit tail rather than the typical holding, so 2 and 4 are judgements, 1 is the floor for a holding that was hit, and 4 is the high case’s average over the holdings hit, not a ceiling on what one farm suffers.

The low and high figures take the low and high end of both assumptions together, so the range spans the assumptions; it is not a margin of error. On Recorded plus fly-tipping on farms, the range allows only for these assumptions: a neighbourhood’s recorded share has its own likely range, shown separately and not added. The farm part counts all fly-tipping on farms, so it includes a very small number of incidents that councils did record on agricultural land; Recorded plus fly-tipping on farms counts those twice, which changes it by well under 1%. The estimate is never added to recorded figures except on that view, where it is labelled as an estimate.

Public land nobody reports

Fly-tipping on public land that nobody reports and no council crew finds cannot be estimated from the data, because nothing measures how much of what the public does not report council crews find. Leeds is the only council whose records say who found each fly-tip, and only for 2018/19 and 2019/20, and those records cannot separate how often the public reports from how much crews find. So the page shows three cases for England as a whole, in words only, if crews find 90%, 70% or 50% of what the public does not report, with the public’s share of records, 0.47, taken from Leeds’s typical ward. Each case scales England’s starting count: recorded fly-tipping, with the 25 councils that record only what the public reports counted as if they also recorded their own staff’s finds, which adds about 3%. They are a sensitivity: none is an estimate, none is added to any total, and none is given for a single neighbourhood or council, because the only evidence on how often the public reports in different places is one council’s 33 wards.

Leeds’s records give only the share of its fly-tip records that came from the public, s = 0.47 at the median ward (April 2018 to March 2020). If its crews found every incident the public did not report, s would be p, the share of incidents the public reports; if they found only a share c of them, s = p / (p + (1 − p)c). So s fixes one combination of p and c, not each, and with c measured nowhere the public-land figure follows from the value chosen for c. Each case applies the same c to Leeds’s crews as to crews everywhere else, so p drops out and the multiplier on the starting count is (1 − s)(1 − c)/c: at c = 0.9, 0.7 and 0.5, 6%, 23% and 53% of the starting count (6%, 23% and 54% of recorded).

Commercial figures

Councils note the main type of waste in each incident, but some waste cannot be identified, and those incidents count as neither construction nor commercial. So every commercial figure is a lower bound, given as “at least X% (Y% if unidentified waste splits like the rest)” and rounded down, so it never overstates the share. If unidentified waste is no more often commercial than the rest, the true share lies between the two. Defra’s construction category also includes large-scale DIY waste from households, so not all of it comes from businesses, and councils classify waste and the size of a load in different ways. The figures show what recorded fly-tipping is made of, not how much fly-tipping businesses do. The shares in the text and in Figure 6 rest on the councils with a full breakdown in each year, which is where unidentified waste is rarest; Defra’s national table, which covers every recorded incident, gives a lower bound of 11.0% for 2024/25 and the same 12.9% if unidentified waste splits like the rest, and both figures are given where the shares appear.

What the factors show

Every factor on this page is an association, not a cause: it shows where to look and what to test, and none shows that changing the factor would change fly-tipping. Within councils, the factors rest on FixMyStreet reports with reporting habits removed; between councils, on councils’ own records (see How councils record). Enforcement also responds to fly-tipping: councils that record more tend to enforce more, so the enforcement figures, in Figure 3 and in a council’s readout, are not what enforcement does. Policy results compare councils before and after a change with other councils over the same years, and none shows that a change made a difference. Drive time to recycling centres shows access only, not a cause of fly-tipping or what closing a centre would do.

Each result is graded by the kind of comparison behind it: within councils, between councils, in the make-up of the waste, or national timing only. Where a 90% range includes no difference, the page says “no detectable association”, because the data cannot tell the difference from none.

Drive times and distances

Drive times are free-flow, with no traffic, over OpenStreetMap roads, from the point where each neighbourhood’s population is centred to the nearest recycling centre of any council, whether or not its residents may use it. Distances to trade and skip-hire sites are straight-line × 1.3, not routed.

Years covered

The map and the national figures are for 2024/25 (April 2024 to March 2025); council comparisons pool it, where stated, with the two years before, and York’s records in the model’s test run to 2025/26. Nothing on this page is a forecast: every fly-tipping figure describes recorded or estimated fly-tipping up to 2025/26.

Neighbourhoods and councils

Neighbourhoods are the Office for National Statistics’ 33,755 small areas for England (2021 lower layer super output areas), each home to roughly 1,000 to 3,000 people. Councils are the 296 English local authorities as they stood in April 2024, and wards are built from whole neighbourhoods (see Wards).

Wards

The map shows 6,793 of England’s electoral wards as at May 2025, the newest for which the Office for National Statistics publishes a lookup from neighbourhoods. Each ward is built from whole neighbourhoods by that best-fit lookup, so where a ward boundary cuts through a neighbourhood the ward shown is an approximation; 406 wards are a single neighbourhood and carry that neighbourhood’s figure, and 33 wards, most of them in the City of London, contain no whole neighbourhood and are not shown. Some councils’ wards changed in May 2026; the map shows the May 2025 wards and the readout says where they have since changed. A ward’s recorded fly-tipping is its neighbourhoods’ estimated shares of the council’s recorded total added together, never a recorded count.

The comparison is one regression across England’s wards of each ward’s rate per household on its deprivation (the English Indices of Deprivation rank, from the 2019 edition, the latest published during 2024/25, as a curve), its population density and the share of its households renting privately (Census 2021), each weighted by households across the ward, with a level of its own for every council, so that each ward is compared with wards in the same council and differences in how councils record do not enter; the decile shown in the readout is from the 2025 edition. Wards in councils whose records carry a recording note, whose conditions lie outside the model’s range, whose 2024/25 total is Defra’s estimate, or where a check on the recorded series is outstanding are shown on the map but not listed in Figure 5. Added 9 October 2026, revised 14 October 2026.