Britain today · Crime & safety

Crime measured by survey has more than halved since 2010 — but fraud, knife offences and police-recorded crime tell different stories

The Crime Survey for England and Wales — the most reliable measure, including unreported crime — shows overall crime well below its 2010 level. Fraud (also survey-estimated) and knife offences (police-recorded, a different method) are the clearest exceptions, both moving the other way.

Read this first — what else changes the meaning of the headline

Overall crime (Crime Survey)
4.33m

Down 55% since 2010 — this is SURVEY-ESTIMATED victimisation, excluding fraud and computer misuse, England and Wales.

ONS, Crime in England and Wales: Appendix Table A1a

Fraud (survey-estimated)
4.07m

Now the single largest crime type, surging since 2017 — measured separately from the "overall crime" figure above, not included in it.

ONS Crime Survey for England and Wales

Knife offences (police-recorded)
48,800

Down 8% from last year but still well above 2011 levels — a POLICE-RECORDED count, a different measurement method from the survey figure above, with two recording-practice breaks in the series (2019, 2024).

ONS, Crime in England and Wales: Appendix Table B6a

Homicide (police-recorded)
499

Down 7% year on year — the lowest since comparable recording began in 2003. Also police-recorded, not survey-estimated (homicide cannot be measured by household survey).

ONS, Crime in England and Wales

4.33m
Crime Survey incidents, excluding fraud and computer misuse
England and Wales · Year ending March 2026 (released 23 July 2026)

55% below the 2010 level on this measure — a genuine long-term fall, though it does not include fraud or computer misuse, both measured separately and both rising.

ImprovingOfficial published data· survey-estimated victimisation, excl. fraudSource: ONS, Crime in England and Wales: Appendix Table A1a

Crime Survey Incidents — England and Wales (millions)

The Crime Survey for England & Wales — the most reliable measure, including unreported crime — shows total crime (excluding fraud) more than halved since 2010. Violent crime and knife crime are the exceptions to this improvement.

View data table
Crime Survey Incidents — England and Wales (millions)England and Wales. Source: ONS, Crime in England and Wales: Appendix Table A1a. Year ending March 2026 (released 23 July 2026).
YearCSEW incidents (m, excl. fraud)
20109.54
20119.71
20129.59
20138.76
20147.55
20157.18
20166.53
20176.06
20186.25
20196.43
20205.66
2021
2022
20234.39
20244.59
20254.6
20264.33

England and Wales · Source: ONS, Crime in England and Wales: Appendix Table A1a · as of Year ending March 2026 (released 23 July 2026)

4.07m
Fraud incidents — the exception moving the other way
Survey-estimated, now the largest single crime type, surged since 2017
Rising
48,800
Knife offences — the second exception
Police-recorded; down from a peak of 55,200, but far above 2011
Down from 2020 peak, still elevated

What the facts show

Crime incidents by type (England and Wales)

Homicide and knife/sharp instrument offences updated to ONS’s Year Ending March 2026 release (published 23 July 2026); remaining categories reflect the prior Year Ending June 2025 release pending refresh. Each row states whether it is survey-estimated (Crime Survey) or police-recorded — the two methods are not directly comparable to each other.

FraudSurvey-estimated↑ Rising
4.07m
Largest crime type — surged since 2017
TheftSurvey-estimated
2.75m
Broadly stable
Violence (without injury)Survey-estimated
1.13m
Far below 2000s peaks
Computer misuseSurvey-estimated↑ Rising
0.66m
Growing rapidly
Criminal damageSurvey-estimated↓ Falling
0.59m
Declining long-term
RobberySurvey-estimated↓ Falling
0.08m
Far lower than 2000s
Knife/sharp instrumentPolice-recorded↓ Falling
48,774
Down 8% — YE March 2026
HomicidePolice-recorded↓ Falling
499
Down 7% from 539 (YE March 2025) — lowest since YE March 2003 recording began

Source: ONS Crime Survey for England and Wales & Police recorded crime — Crime in England and Wales: year ending March 2026 (homicide, knife/sharp instrument) and year ending June 2025 (other categories, pending refresh).

Knife and Sharp Instrument Offences — England and Wales ('000s)

Knife and sharp instrument offences have risen steadily since 2014, concentrated among young men in deprived urban areas — the clearest police-recorded exception to the general Crime Survey decline.

View data table
Knife and Sharp Instrument Offences — England and Wales ('000s)England and Wales. Source: ONS, Crime in England and Wales: Appendix Table B6a (including Greater Manchester). Year ending March 2026 (released 23 July 2026).
YearKnife offences ('000s)
201136.1
201233.7
201328.9
201428.3
201529.2
201632.7
201739.5
201847.5
201952.5
202055.2
202144.7
202249.2
202351.7
202454
202553.1
202648.8

England and Wales · Source: ONS, Crime in England and Wales: Appendix Table B6a (including Greater Manchester) · as of Year ending March 2026 (released 23 July 2026)

Policing

Police officer numbers were significantly cut after 2010, partially recovered under the 2019–2023 uplift programme, but a precise per-capita (officers per 1,000 population) series could not be independently confirmed against matching ONS population denominators. The confirmed absolute headcounts (Home Office police workforce statistics, full-time equivalent) are shown below. Meanwhile, crime has become more complex — particularly fraud, cybercrime, and organised crime — stretching resources further.

141,690
Police officers (FTE)
2010 — Pre-austerity
119,959
Police officers (FTE)
2018 — Post-austerity low
146,868
Police officers (FTE)
Sep 2024 — Post-uplift programme

Source: Home Office, Police workforce, England and Wales. Absolute headcounts; a per-1,000-population ratio was not shown as it could not be confirmed against matching ONS mid-year population estimates.

“Crime has fallen” is true and incomplete at the same time

Different crime types, measured different ways, over different periods, do not add up to one simple story.

Two different measurement methods, not directly comparable

The Crime Survey for England and Wales (CSEW) estimates victimisation by asking ~45,000 households about their experience, including crime never reported to police. Police-recorded crime counts only what was reported and recorded. They can move in different directions in the same year for the same offence type, and neither is simply "more accurate" than the other — they measure different things.

Source: ONS Crime Survey for England and Wales; Home Office Police Recorded Crime

Fraud and computer misuse are excluded from the headline "overall crime" figure

The 4.33m CSEW figure this page leads with excludes fraud and computer misuse — the only basis that runs continuously back to 2010, because those categories were added to the survey later. Fraud alone (4.07m) is now nearly as large as the entire "overall crime" total shown as improving — a reader who saw only the headline could reasonably think fraud was included in the decline. It is not.

Source: ONS, Crime in England and Wales: Appendix Table A1a

The knife-offences series has two recording breaks that limit comparability

Recording practices changed in April 2019 (NDQIS) and the Metropolitan Police corrected its own recording in April 2024 without revising earlier years. Comparing knife-offence figures across either break point is not a clean like-for-like comparison, even though the series is presented as one continuous line for readability.

Source: ONS, Crime in England and Wales: Appendix Table B6a

There is a genuine data gap, not just a low point, in 2021–2022

Face-to-face CSEW interviewing was suspended during COVID, so ONS published no comparable figures for the years ending March 2021 and 2022. The chart above shows a real gap, not a dip to zero — treat any "crime fell during COVID" claim built from this series with that gap in mind.

Source: ONS Crime Survey for England and Wales

What it means

What the data directly shows
CSEW-measured crime (excluding fraud and computer misuse) has fallen by more than half since 2010; fraud and knife offences, measured separately and by different methods, have both risen over a similar period; homicide is at its lowest recorded level since 2003.
What can reasonably be inferred
Britain is, on the best available continuous survey measure, safer from the crime types that measure covers than it was in the early 2010s — but 'crime' as a single word covers categories moving in opposite directions, so a single 'safer' or 'more dangerous' verdict for the whole picture would overstate what the evidence shows.
What is disputed
How much of the shift toward fraud reflects genuinely more fraud happening, versus better detection and reporting as more of life moved online, is disputed among criminologists — the data on this page cannot fully separate the two explanations.
A political judgement, not a finding
Whether policing resources should shift toward fraud/cybercrime investigation, toward knife-crime prevention, or stay distributed as they are is a political and operational choice this data does not make for the reader.
What the evidence cannot establish
This data cannot establish the true scale of unreported, undetected crime by definition, nor precisely how much of the knife-offences trend is a real change in offending versus a change in recording practice given the two documented breaks in that series.

What people often get wrong

What people often think

Crime has fallen, so Britain is simply safer than it was.

What the evidence shows

Overall Crime-Survey-measured crime (excluding fraud) has fallen sharply — but fraud, now the single largest crime type, and knife offences have both risen over a similar period. 'Safer' depends entirely on which crime type and which period you mean.

Source: ONS Crime Survey for England and Wales

What can change this

Where there are meaningful levers, drivers or constraints on this picture.

Specialist digital/fraud investigation capacity

Current policy

Fraud requires specialist digital investigators — a different skill set and resourcing model from traditional policing, and the fastest-growing crime category on this page.

Evidence:
Fraud incidents (4.07m, survey-estimated) are now the largest single crime category, as shown above.
Expected effect:
Increased specialist capacity would plausibly improve detection and prevention rates, though this page does not carry a registry-verified figure quantifying that effect.
Time horizon:
Years — specialist training and capability-building take time.
Trade-offs:
Resources moved toward fraud investigation are resources not available for other policing priorities, given headcount only recently recovered to 2010 levels.
Unknowns:
How much of the fraud rise reflects genuinely more offending versus better detection as more of life moved online.

Continue police workforce levels at or above the post-uplift level

Current policy

Police officer numbers fell from 141,690 (2010) to a low of 119,959 (2018) before the 2019-2023 uplift programme brought them to 146,868 (Sep 2024).

Evidence:
Home Office police workforce statistics, as tabled above.
Expected effect:
Sustained workforce levels support capacity across all crime types, including the more complex categories (fraud, cybercrime, organised crime) that require more resource per case.
Time horizon:
Ongoing — officer training and retention take years.
Trade-offs:
Requires sustained budget commitment against other public-spending priorities.
Unknowns:
The precise relationship between officer numbers and crime rates is contested in the wider research literature and not settled by the figures on this page alone.

Targeted knife-crime prevention in the areas most affected

External option

Knife offences are concentrated among young men in deprived urban areas — a targeted rather than universal approach is what most existing prevention programmes attempt.

Evidence:
Knife and sharp instrument offences chart above; concentration pattern per ONS commentary.
Expected effect:
Targeted programmes have shown localised effects in some areas, though this page does not carry a registry-verified national evaluation figure.
Time horizon:
Years for a sustained trend change; some interventions show effects sooner.
Trade-offs:
Targeting specific areas/demographics raises its own fairness and stigmatisation questions that are outside this page's evidence base.
Unknowns:
How much of the recent 8% fall in knife offences reflects any specific policy versus the two recording-practice breaks documented above.

Local comparison

How this plays out locally

The figures above are England-and-Wales-wide totals from the Crime Survey and national police-recorded statistics. Hampshire’s local authority pages carry their own force-level crime module (Hampshire Constabulary data, a different geography and reporting basis from the national figures here) rather than a number recomputed from this page’s totals.

See Portsmouth’s local crime picture →

Detail data and technical notes

Overall crime: ONS, Crime in England and Wales: Appendix Table A1a. CSEW headline crime EXCLUDING fraud and computer misuse — the only basis that runs continuously from 2010. Crime more than halved, from 9.5m in the year to March 2010 to 4.33m in the year to March 2026. NOTE THE GAP: face-to-face interviewing was suspended for COVID, so ONS published no figures for the years ending March 2021 and 2022.

Knife offences: ONS, Crime in England and Wales: Appendix Table B6a (including Greater Manchester). Police-recorded offences involving a knife or sharp instrument, including homicide, years ending March. Offences rose steeply from 2014 to a 2020 peak, and fell 8% in the year to March 2026 to 48,774. TWO BREAKS IN SERIES: recording changed in April 2019 (NDQIS) and the Metropolitan Police corrected its recording in April 2024 without revising earlier years — comparisons across either break are not like-for-like.

Fraud, theft, violence, computer misuse, criminal damage and robbery figures reflect the prior Year Ending June 2025 CSEW release, pending refresh to the Year Ending March 2026 basis used for homicide and knife offences above — the two release dates are stated explicitly on each figure rather than implied to be the same.