Crime at UK Schools: What 39 Police Forces Revealed in a Freedom of Information Investigation

A Teneo International School FOI investigation of 39 UK police forces found 96,559 offences at school premises over three years across 23 forces providing full data. Overall crime fell 15% — but reported sexual offences rose in 11 forces and weapon offences in five, against that trend. Sixteen forces refused on Section 12 cost grounds.

Between July and September 2026, Teneo International School submitted Freedom of Information requests to every UK territorial police force. Thirty-nine forces engaged. Twenty-three provided full data across all four questions, or close enough to it to be usable. Sixteen refused on Section 12 (cost limit) grounds in whole or in part. The 96,559 school-premises offences recorded across the three-year reference period make this the largest single public investigation into police-recorded crime at UK schools — and the picture it draws is sharper, more uneven, and more revealing of recording practice itself than any single headline captures.

This is the full dataset, force by force, with the caveats and structural findings that come with it.

What the investigation asked

Each force received an identical four-question request:

• Whether the force can isolate offences recorded at school premises through a location flag or similar identifier.

• Total offences recorded at school premises over the three most recent available years, broken down by offence type.

• Assaults on school staff recorded at school premises, over the same period.

• Possession offences involving bladed articles and other offensive weapons at school premises, over the same period.

Only aggregate numbers were requested — no personal data, no information identifying specific schools, pupils or staff. Three years of data was requested in either financial-year or calendar-year format depending on how each force held its records.

Headline findings

Scale:

• 39 UK territorial police forces engaged with the request.

• 23 provided full data on offence totals.

• 3 provided weapon-only data (Gwent, Wiltshire, Avon & Somerset — Section 12 on the other questions).

• 13 refused on Section 12 cost grounds across the entire request (blanket refusals).

• 96,559 offences recorded at school premises across three years in the 23 full-data forces.

Overall trend:

• Total offences at schools fell 15% over three years across the 23 providing forces (34,482 → 29,280).

• Sharpest single-force drops: Essex (−34%), Dorset (−33%), Northamptonshire (−29%), Norfolk (−26%), Cleveland (−24%), Cumbria (−20%), Leicestershire (−19%).

• Dyfed-Powys was the outlier — total offences at schools rose 14% (627 → 715), the only clear rising-total force in the data.

Sexual offences — the strongest rising pattern:

• Reported sexual offences at schools rose in 11 forces of 23 providing data.

• Sharpest rises: Dyfed-Powys +84%, Nottinghamshire +50%, Devon & Cornwall +49%, Greater Manchester +31%, Northamptonshire +24%, Durham +18%, Suffolk +18%, Leicestershire +15%, Merseyside +14%, Norfolk +7%, Cleveland +2%.

• Flat or falling in: Sussex, Staffordshire, West Mercia, Dorset, Lancashire, Essex, South Wales, West Midlands, Cumbria.

Weapon offences — a secondary rising pattern:

• Weapon offences at schools rose in five forces: Greater Manchester Police (blade/point +38%), Avon & Somerset (+29%), Durham (+23%), South Wales (+15%), Essex (+41% in the latest year).

• Dyfed-Powys +108% on a small base (12 → 25).

• Cleveland Police fell sharpest, −62% (13 → 5).

• Most other forces broadly flat.

Assaults on school staff — the recording gap:

• Only eight forces could provide clean data on staff-victim assaults.

• Leicestershire Police recorded 251 staff assaults across three years — the highest in the dataset — with 91 in a single year.

• South Yorkshire Police: 82, 72, 62 teacher-victim assaults over three years.

• West Midlands Police: 951 violence-against-the-person offences at schools in 2023 — only 2 recorded with a teacher victim. The single sharpest illustration of the occupation-field recording gap.

• Cumbria: provided staff numbers, but 88% of relevant records had no occupation field completed.

• Devon & Cornwall: 9 confirmed staff assaults over three years, with 2,085 unassessed assault records where occupation was not recorded.

The scale of the investigation

Forces providing full offence-total data (23): Avon & Somerset Police (weapons only), Cleveland Police, City of London Police, Cumbria Constabulary, Devon & Cornwall Police, Dorset Police, Durham Constabulary, Dyfed-Powys Police, Essex Police, Greater Manchester Police, Lancashire Constabulary, Leicestershire Police, Merseyside Police, Norfolk Constabulary, North Wales Police, Northamptonshire Police, Nottinghamshire Police, South Wales Police, South Yorkshire Police, Staffordshire Police, Suffolk Constabulary, Sussex Police, West Mercia Police, West Midlands Police.

Forces providing partial data (weapons only; Sections 12 on offence totals and staff assaults): Gwent Police, Wiltshire Police, Avon & Somerset Police.

Forces refusing in full on Section 12 cost grounds (13): Police Service of Northern Ireland (PSNI), West Yorkshire Police, Hampshire & Isle of Wight Constabulary, Derbyshire Constabulary, Police Scotland, Metropolitan Police Service (MPS), North Yorkshire Police, Cheshire Constabulary, Thames Valley Police, Kent Police, Humberside Police, Gloucestershire Constabulary, Northumbria Police.

On the Section 12 refusals. The pattern of refusals is not uniformly explained by force size or recording infrastructure. Several neighbouring forces of comparable size made different decisions:

• Essex Police provided a detailed breakdown. Kent Police — adjoining Essex, sharing FOI infrastructure per Kent's own response, with comparable population — refused on Section 12 grounds.

• Norfolk and Suffolk issued a joint response providing full data. Several similarly-sized East of England forces refused.

• South Wales Police and Gwent Police both engaged (Gwent on weapons; South Wales on everything), demonstrating the data is extractable.

Section 12 refusals are decisions made within the Freedom of Information Act's discretion. The paired-infrastructure examples demonstrate that the data is extractable within statutory cost limits by forces that choose to prioritise doing so.

Reported sexual offences at schools: the strongest finding

The single most consistent cross-force pattern in the investigation is a rise in reported sexual offences at school premises. Eleven of the 23 forces providing comparable data recorded rises over the three-year reference period, several of them substantial. The pattern holds even where overall school crime fell — a rise against a falling trend.

Specific force-by-force figures:

• Dyfed-Powys Police: 58 → 80 → 107 — a +84% rise, the sharpest single-force rise in the entire dataset.

• Nottinghamshire Police: 78 → 116 → 117 — +50% on the base year.

• Devon & Cornwall Police: combined sexual offences 127 → 189 over three years — +49%. Driven by Other Sexual Offences rising from 107 to 180 (+68%); Rape fell from 20 to 9.

• Greater Manchester Police: 293 → 357 → 385 — +31% over three years, in the largest-population force outside London.

• Northamptonshire Police: 143 → 182 → 177 — +24% on the base year, peaking in 2024.

• Durham Constabulary: 61 → 74 → 72 — +18% on the base year.

• Suffolk Constabulary: 82 → 97 (two years comparable) — +18%, against an 18% fall in overall offences at Suffolk schools.

• Leicestershire Police: 107 → 131 → 123 — +15% on the base year.

• Merseyside Police: 134 → 178 → 153 — +14% on the base year, with 2024 the peak.

• Norfolk Constabulary: 98 → 92 → 105 — +7% on the base year, against a 26% fall in overall school offences.

• Cleveland Police: 124 → 81 → 126 — +2% on the base year, with a dip in 2024.

Flat or falling sexual offences:

• Sussex Police (likely partial-year effect in latest figures).

• Staffordshire, Dorset, Lancashire, West Midlands all broadly flat.

• West Mercia Police: peaked 2024 (208) then fell to 184.

• Essex Police: 130 → 129 → 118 — fell overall.

• South Wales Police: 156 → 172 → 138 — fell overall.

• Cumbria Constabulary: 48 → 86 → 47 — peaked in 2024 then returned to base.

Essential context on sexual offence reporting

A rise in reported offences does not necessarily indicate a rise in actual offending. The pattern is consistent with — and in the view of child-protection specialists most likely reflects — more young people coming forward and reporting incidents that would previously have gone unreported. This reflects sustained safeguarding work, improved teacher training on disclosure, and public campaigns on sexual violence.

The data is also sensitive. A proportion of recorded offences involve incidents between young people. Figures are offences recorded with a school-premises marker, which does not always indicate the offence occurred during school hours or involved a pupil of that school.

Support available:

• NSPCC — 0808 800 5000

• Childline — 0800 1111

Weapon offences at schools: a secondary rising pattern

Weapon offences at school premises rose in five forces over the three-year period, running against the generally falling trend in overall school crime.

Forces where weapon offences rose:

• Greater Manchester Police: blade/point possession at schools 79 → 97 → 109 across three years — a +38% rise, the sharpest among larger forces (GMP covers the largest population outside London).

• Avon & Somerset Police: total weapon offences at schools 42 → 47 → 54 — +29% over three years. Rises across blade/point (30 → 28), other offensive weapons (5 → 12), threat with blade (7 → 10), and threat with weapon (0 → 4).

• Durham Constabulary: total weapons 13 → 14 → 16 — +23% over three years. Driven by a sharp rise in "other offensive weapon" from 1 to 7 in the latest year.

• South Wales Police: total weapon offences at schools 33 → 28 → 38 — +15% on the base year, against a 10% fall in overall school crime.

• Essex Police: weapon offences at schools 22 → 22 → 31 — a late rise of 41% in the latest year, against the force's 34% fall in overall school crime (the sharpest overall drop in the dataset).

• Dyfed-Powys Police: weapon offences 12 → 12 → 25 — +108% over three years on a small base.

Forces where weapon offences fell:

• Cleveland Police: 13 → 8 → 5 — a 62% fall, running sharply counter to the national pattern.

• West Mercia Police: 31 → 29 → 26 — a 16% fall.

• Sussex Police: 63 → 59 → 44 — a 30% fall (likely partial-year effect in latest figures).

Forces where weapon offences were broadly flat or peaked-then-fell:

• West Midlands Police: 220 → 174 → 197 — the highest absolute weapon numbers in the dataset, with 371 blade/point offences across three years, materially larger than any other single force.

• South Yorkshire Police: 49 → 37 → 54 — recovered to base.

• Lancashire Constabulary: 13 → 35 → 30 — peaked 2024, fell back in 2025.

• Norfolk Constabulary: 27 → 38 → 12 — sharp 2025 drop, possibly partial year.

• North Wales Police: 4 → 13 → 8 — peaked 2024.

• Gwent Police: 22 → 28 → 20.

• Nottinghamshire Police: 18 → 9 → 17 — dipped then recovered.

• Cumbria Constabulary, Leicestershire Police, Suffolk Constabulary: all broadly flat.

Welsh picture on weapons specifically: across the three Welsh forces providing weapon data in the latest year — North Wales (8), South Wales (38), Gwent (20) — a combined 66 weapon offences at Welsh schools. Dyfed-Powys adds another 25, making 91 across all four Welsh forces.

Across all 24 forces providing weapon data, the three-year aggregate was 782 → 789 → 806. The aggregate trend is marginally upward (+3%), but driven heavily by the five rising forces offsetting flat or falling results elsewhere.

Assaults on school staff: a recording gap, not a data vacuum

The investigation asked each force for the number of assaults where the victim was recorded as a school staff member. Only eight forces could provide clean data to this question. This is not because the assaults do not happen — it is because of a national recording gap in how police forces capture the occupation of victims.

Forces providing clean staff-victim assault data:

• Leicestershire Police: 91, 73, 87 staff assaults — 251 across three years — the highest total in the dataset. Includes an unusual breakdown: 11, 4, 8 of these involved a weapon. Leicestershire separately recorded 2 sexual assaults on staff in 2024 and 2025.

• Nottinghamshire Police: 25, 46, 21 — 92 across three years.

• Essex Police: 30, 16, 16 — 62 across three years. Teachers dominant (29/14/15); School Secretary 1/1/0; Classroom Assistant 0/1/0; Sports Instructor 0/0/1.

• Cumbria Constabulary: 25, 14, 21 — 60 across three years. Breakdown: 33 Education Teacher, 14 Teacher, 7 Education, 3 Classroom Assistant, 2 School Secretary, 1 Caretaker. But 88% of relevant records had no occupation field completed — the actual total is almost certainly materially higher.

• Dorset Police: 24, 22, 7 — 53 across three years. Breakdown: 22 Teacher, 1 Caretaker, 1 Secretary. Sharp fall over the period, requiring interpretation (real drop or recording-practice change).

• North Wales Police: 12, 13, 11 — 36 across three years. Detailed incident list provided: mostly common assault, some ABH, harassment. Teachers dominant, classroom assistants and caretaker also recorded as victims.

• West Midlands Police: 2, 2, 0 — 4 across three years. Set against 2,811 violence-against-the-person offences at West Midlands schools across the same period. The ratio illustrates the recording gap in its sharpest form.

• City of London Police: 0, 0, 0 — nil values, credible given force geography (financial square mile).

Devon & Cornwall Police provided a partial figure: 1, 4, 4 — 9 across three years. With the explicit caveat that 2,085 assault records in the same period had no occupation recorded — meaning the 9 is a floor against a potential ceiling two orders of magnitude higher.

Why the recording gap matters

Forces that refused the staff-assault question on Section 12 grounds typically cited the volume of manual review required to establish victim occupation. The specific numbers reveal the scale:

• North Yorkshire Police: 900+ assault reports would need manual review.

• Northumbria Police: 6,812 records at 5 minutes each = 567 hours.

• Kent Police: 1,000+ VwI reports.

• Humberside Police: 1,411 school-premises crimes × 15 min = 352 hours.

• Durham Constabulary: 2,000 records to review.

• MPS: 150+ records for CRIS alone, 6,000+ in scope across systems.

• Merseyside Police: review would exceed the 18-hour statutory limit.

• Staffordshire Police: 1,100+ VAP records would need manual check.

Wiltshire Police's structural admission is the sharpest in the entire dataset: "The reason this field is not mandatory is due to the fact there is no statutory requirement for Police Forces to record this data — it is purely arbitrary."

Hampshire & Isle of Wight Constabulary stated the position in parallel terms: "no centrally searchable marker to identify whether an offence occurred at school premises."

The practical implication is unambiguous: the number of assaults recorded as having a school staff victim is a floor, not a ceiling. The actual national figure is almost certainly multiples of what forces have been able to disclose. The eight forces that could answer the question demonstrate it is answerable; the Section 12 refusals cite scale, not impossibility.

Overall offences at schools: the falling trend

Alongside the specific category-level patterns above, the broader picture is of total recorded crime at schools falling in most forces providing data.

Three-year falls (base year to latest year):

• Essex Police: 2,013 → 1,322 — −34% (sharpest in dataset).

• Dorset Police: 1,204 → 809 — −33%.

• Northamptonshire Police: 1,426 → 1,008 — −29%.

• Norfolk Constabulary: 1,081 → 799 — −26%.

• Cleveland Police: 1,208 → 912 — −24%.

• Cumbria Constabulary: 678 → 542 — −20%.

• Leicestershire Police: 1,297 → 1,056 — −19%.

• Suffolk Constabulary: 884 → 727 — −18%.

• West Mercia Police: 1,710 → 1,418 — −17%.

• Merseyside Police: 1,891 → 1,577 — −17%.

• Staffordshire Police: 1,548 → 1,276 — −18%.

• Durham Constabulary: 675 → 581 — −14%.

• South Yorkshire Police: 2,841 → 2,480 — −13%.

• Lancashire Constabulary: 2,421 → 2,225 — −8%.

• South Wales Police: 1,692 → 1,524 — −10%.

• Greater Manchester Police: 5,048 → 4,728 — −6%.

• West Midlands Police: 1,698 → 1,669 — −2% (broadly flat).

Nottinghamshire Police: 991 → 1,166 → 1,017 — rose then fell, net +3%.

Dyfed-Powys Police: 627 → 715 — +14%, the only clear rising-total force in the dataset.

Violence against the person was the single largest offence category in nearly every force providing a breakdown — typically between 46% and 58% of all school-premises offences. Public order, sexual offences, criminal damage, and stalking & harassment made up most of the remainder in varying proportions.

Partial-year caveats

Several 2025 and 2025/26 figures may reflect partial reporting cycles rather than full years:

• Sussex Police: 2025 figure of 994 is roughly half the 2024 figure (1,854), suggesting an 8-month cut rather than a full 2025 year.

• Norfolk Constabulary: 2025/26 weapons dropped sharply (38 → 12), likely partial year.

• Lancashire Constabulary: provided 2026 YTD alongside full years — partial-year basis flagged.

• Dyfed-Powys Police: 2025/26 reported alongside YTD 2026 figure of 363 — some latest-year figures may be partial.

Where year-on-year comparison is drawn above, we have excluded partial-year latest figures where the force flagged this. Full-year verification would adjust some of the comparisons.

Section 12 refusals: a choice, not a necessity

Sixteen forces refused the request on Section 12 cost grounds in whole or in part. Thirteen refused across all four questions; three (Gwent, Wiltshire, Avon & Somerset) provided weapon data but refused the broader request.

The structural pattern across refusals:

• Forces could typically search by Home Office code (which enables weapon-offence counts), but not by location flag (which would enable school-premises totals).

• Occupation of victim is a non-mandatory field in nearly every force's recording system, which prevents staff-victim identification without manual review.

• Several forces noted that the "schools" location marker, where it exists, often combines schools with nurseries, colleges, or further-education premises.

• Section 12 refusals are triggered by the staff-assault question in some forces, which then cascades to a blanket refusal of the entire request.

The sharpest structural admissions:

• Wiltshire Police: "The reason this field is not mandatory is due to the fact there is no statutory requirement for Police Forces to record this data — it is purely arbitrary."

• Hampshire & Isle of Wight Constabulary: "no centrally searchable marker to identify whether an offence occurred at school premises."

• MPS: pre-February 2024 records cannot separate schools from nurseries without manual review of each investigation. The newer CONNECT system has separated codes, but legacy data cannot be back-separated.

• Police Scotland: cannot search by locus at all; staff-victim occupation searchable only from January 2024 and non-mandatory.

• Northumbria Police: "This information is not held centrally nor is it held in a format that allows its extraction."

Scale of what Section 12 refusals represent:

• West Yorkshire Police: 6,857 VAP crimes recorded with "school" location qualifier — one force.

• MPS: 6,000+ records in scope.

• Northumbria Police: 6,812 records.

• Derbyshire Constabulary: 3,474 records with "school" in address.

• Humberside Police: 1,411 school-premises crimes across three years.

Had the data been releasable from these forces, the national three-year aggregate would be materially higher than 96,559.

The best-recorded forces: a case study in what's possible

Three forces stand out for the completeness and granularity of their responses.

Leicestershire Police provided the best-quality data structure in the entire sample. Full staff-vs-non-staff split for every offence category; weapon-involvement flag for staff-victim assaults; detailed category-level breakdowns across three years. This is the dataset structure the Freedom of Information Act was designed to enable — and demonstrates that recording-and-extraction limitations elsewhere are not technological inevitabilities.

North Wales Police provided detailed per-incident lists for staff assaults and weapons offences, with cross-referenced HO codes allowing independent verification of classification. The force's response is the cleanest single dataset in the sample.

Dorset Police provided a 14-category breakdown with explicit flagging of the "offences recorded at schools versus occurring elsewhere" question — a nuance most forces didn't surface — and became the second force (after North Wales) to provide clean staff-victim figures.

These three forces together demonstrate that the questions are answerable, the data is extractable, and the recording infrastructure exists. The variation across the sample reflects force-level choices about recording practice, FOI resourcing, and prioritisation of transparency.

What the investigation does not — and cannot — tell us

Being direct about the limits of what this data can show:

• Causation. The data records offences reported to police. It does not reveal why rates in a specific category are rising in one force and falling in another.

• Severity. An offence record captures that an offence occurred; it does not uniformly capture severity, injury level, or outcome.

• Reporting rates vs incident rates. A rise in reports is not the same as a rise in actual offending. This is particularly important for sexual offences, where improved reporting is a public policy goal.

• School type. Figures cover all school premises recorded by the force, without consistently distinguishing between primary, secondary, special school, or further-education premises. Several forces flagged difficulty separating school from further-education premises.

• Time of day. A school-premises marker does not confirm an incident occurred during school hours or involved pupils.

• School-of-record vs incident location. Several forces (notably Dorset) flagged that an offence may be recorded with a school as its occurrence address when the actual incident took place elsewhere.

• Specific schools. Figures are aggregate at force level. No individual school is identified or identifiable.

Each limitation above was flagged by at least one responding force in its response letter. We have reflected them throughout the findings.

Methodology

Request structure. Standardised FOI request sent to all UK territorial police forces across England, Wales, Scotland, and Northern Ireland. Each force received an identical four-question request.

Reference period. The three most recent available years at time of request. Forces responded in calendar-year format (2023–2025) or financial-year format (2022/23–2024/25) depending on how the force holds its records. Comparisons across forces are made on the three-year reference each force provided, not forced to a common year basis.

Data treatment. As released by each force, with no secondary analysis beyond addition, percentage change, and categorisation into themes. Where a force flagged a specific caveat (occupation field completeness, school vs further-education premises, redacted sub-5 figures, partial-year reporting), the caveat is reflected in our use of that data.

Aggregation. The 96,559 three-year total represents the sum of total school-premises offences from the 23 forces providing full offence-total data. This is a floor on the UK national figure. Approximately 16 UK territorial forces (counting blanket and partial S12 refusals) are not represented in the aggregate — including some of the largest forces by population (MPS, West Yorkshire Police, Police Scotland, PSNI).

Verification. All numeric figures published are drawn directly from force FOI responses, each identified by FOI reference and response date in the underlying tracker. Response letters are available on request to media and researchers.

Known partial-year caveats. 2025 and 2025/26 figures from Sussex Police, Norfolk Constabulary, Lancashire Constabulary, and Dyfed-Powys Police may reflect partial reporting cycles rather than full years. Comparisons involving these forces' latest years should be treated as indicative pending full-year confirmation.

Full data available

The full FOI dataset — including force-by-force breakdowns, response letters, FOI references, and the standardised four-question request text — is available on request to journalists, researchers, and policy organisations.

The dataset is published to be used. Journalists are welcome to quote the figures with attribution to Teneo International School; policy organisations and academics are welcome to request the underlying data for further analysis.

Commentary

Taryn Jankes, Chief Marketing Officer of Teneo International School, said:

"When overall crime at schools is falling by 15% across the forces we could count, but reported sexual offences at schools are rising in eleven forces and weapon offences in five, the picture is more complicated than any single headline captures. The falling totals are real and worth noting. So are the specific categories moving in the opposite direction.

"The sexual-offences pattern most likely reflects more young people coming forward and reporting incidents that would previously have gone unreported. That is a necessary and welcome shift — it is also a demand on the services that take those disclosures seriously. The weapons picture is less uniform, but five forces moving the same direction against the overall trend is a pattern that deserves attention.

"The deeper finding of this investigation is what the data cannot tell us. Only eight forces could isolate assaults where the victim was a member of school staff. West Midlands Police recorded 951 violent offences at its schools in a single year with 2 having a teacher as recorded victim; Devon & Cornwall confirmed 9 staff assaults against 2,085 unassessed assault records; Cumbria reported staff-assault figures with 88% of the relevant records missing occupation data; Wiltshire Police put the position plainly in its own response — the field is not mandatory because recording this data is 'purely arbitrary'. The occupation gap in police recording is a national issue, not a local one. These figures are the floor, not the ceiling."

Support

• NSPCC — 0808 800 5000 — if you are concerned about a child's welfare.

• Childline — 0800 1111 — free support for young people under 19.

• National Education Union / NASUWT — support for school staff who have experienced assault or threat at work.

About the investigation

The investigation was conducted by Teneo International School between July and September 2026. All FOI requests were submitted through official force channels using Teneo International School's standard FOI request template.

Teneo International School is a British International online school for learners aged 11 to 19, with 20,000+ learners, 4,000+ graduates, and 170+ teachers worldwide. The school's interest in the investigation reflects its focus on safeguarding and school environment as educational priorities.