Crime & safety · Serious violence · England and Wales · the year to March 2026

Is knife crime and serious violence rising or falling?

Police-recorded knife offences fell 8.1% (YE Mar 2026). Homicide: 499 recorded deaths in 2025–26. Sharp-object assault admissions: 3,337 in 2025–26 (provisional). These measure different harms; recording changes limit long comparisons.

Evidence last reviewed 9 October 2026

Knife offences recorded by the police

Is knife crime going up or down?

Police-recorded knife offences fell 8.1% (YE Mar 2026)

Selected violent and sexual offences involving a knife or sharp instrument recorded by the police, including homicide, thousands, 2010–11 to 2025–26, in two recording runs.

England and Wales (including Greater Manchester and British Transport Police) · 2010–11 to 2025–26

1996–2026 displayed window · observations only. Shading marks years outside the displayed record.

Police-recorded knife offences fell 8.1% (YE Mar 2026). The chart separates the recording methods. Improvements in identification and recording mean the full historical rise cannot be treated as a like-for-like change in offending.

England and Wales (including Greater Manchester and British Transport Police) · 2010–11 to 2025–26 · Official statistics (not accredited)

Sources, definitions and data

Geography:: England and Wales (including Greater Manchester and British Transport Police) — All recorded offences, victims of any age
Measure:: Thousand offences
Period:: 2010–11 to 2025–26
Status:: Official statistics (not accredited)
Calculation:: As published (Worksheet B6, “total selected offences including homicide (inc GMP)”). Split into two runs at the April 2019 method change.
Compare with:: 2024–25 (53,096) and 2019–20 (55,170), both on the NDQIS basis
Not included / breaks:: Before April 2019 most forces improved how they identify knife offences, so earlier years are not directly comparable. Greater Manchester under-counted before December 2017. The Metropolitan Police corrected its recording from April 2024 without revising earlier years. Possession offences are not included.

Source: ONS, Crime in England and Wales: Appendix tables, year ending March 2026, Worksheet B6 (including Greater Manchester) (released 23 July 2026). Chart CRIME-08.

View the data as a table

Year	Offences	Recording basis
2010–11	36,068	Before NDQIS
2011–12	33,740	Before NDQIS
2012–13	28,902	Before NDQIS
2013–14	28,337	Before NDQIS
2014–15	29,161	Before NDQIS
2015–16	32,724	Before NDQIS
2016–17	39,468	Before NDQIS
2017–18	47,467	Before NDQIS
2018–19	52,511	Before NDQIS
2019–20	55,170	NDQIS
2020–21	44,728	NDQIS
2021–22	49,189	NDQIS
2022–23	51,741	NDQIS
2023–24	53,995	NDQIS
2024–25	53,096	NDQIS
2025–26	48,774	NDQIS

Police-recorded knife offences fell 8.1% (YE Mar 2026). The first observation shown is 36,068 in 2010–11 — but forces changed how they identify knife crime in between, so that comparison mixes real change with better recording.

Knife crime is concentrated. The Metropolitan Police recorded 28% of offences in the year to March 2026, and most take place in metropolitan areas. Most are robberies or assaults with injury; homicide is under 1% of the total. Use the area explorer for local recorded figures.

Homicide

How many homicides?

499 homicides: the fewest since current recording began in 2002–03

Homicide offences recorded by the police (murder, manslaughter, corporate manslaughter, infanticide), England and Wales, 2002–03 to 2025–26.

England and Wales · 2002–03 to 2025–26

1996–2026 displayed window · observations only. Shading marks years outside the displayed record.

Homicide is rare enough to be counted almost completely, so recording changes matter less than for other violence. Small numbers and major incidents can move the line. The July 2026 ONS bulletin reported 176 knife homicides for 2025–26, the fewest since 1990–91 on the Homicide Index; that is a dated bulletin finding, not an automatically refreshed breakdown.

England and Wales · 2002–03 to 2025–26 · Official statistics (not accredited); recent years are revised as cases go through the courts

Sources, definitions and data

Geography:: England and Wales
Measure:: Offences
Period:: 2002–03 to 2025–26
Status:: Official statistics (not accredited); recent years are revised as cases go through the courts
Calculation:: As published (Table A5a).
Compare with:: 2024–25 (539)
Not included / breaks:: Some years include single events recorded at once: 173 Shipman victims in 2002–03, 52 victims of the 7 July bombings in 2005–06, 96 Hillsborough victims in 2016–17, 35 victims of the 2017 attacks in 2017–18 and 39 people found in a lorry in 2019–20.

Source: ONS, Crime in England and Wales: Appendix tables, year ending March 2026, Table A5a (released 23 July 2026). Chart CRIME-09.

View the data as a table

Year	Homicides
2002–03	1,047
2003–04	904
2004–05	868
2005–06	764
2006–07	758
2007–08	775
2008–09	664
2009–10	620
2010–11	639
2011–12	553
2012–13	558
2013–14	533
2014–15	539
2015–16	577
2016–17	709
2017–18	714
2018–19	673
2019–20	711
2020–21	573
2021–22	687
2022–23	581
2023–24	567
2024–25	539
2025–26	499

A measure police recording cannot move

What do hospitals see?

Hospital admissions for sharp-object assault are 36% below their 2018–19 peak

Admissions to NHS hospitals in England and Wales for assault by a sharp object, all ages, 2012–13 to 2025–26.

England and Wales (NHS hospitals) · 2012–13 to 2025–26

1996–2026 displayed window · observations only. Shading marks years outside the displayed record.

This measure does not depend on whether a crime was reported or how police record it, which is why the Home Office used it to evaluate violence-reduction funding. It tells a similar story to recorded knife offences: a rise to the late 2010s and lower levels since. The latest year is provisional and may change when late records arrive.

England and Wales (NHS hospitals) · 2012–13 to 2025–26 · Official statistics; 2025–26 is provisional and likely to be revised up

Sources, definitions and data

Geography:: England and Wales (NHS hospitals) — Inpatients of all ages
Measure:: Finished admission episodes
Period:: 2012–13 to 2025–26
Status:: Official statistics; 2025–26 is provisional and likely to be revised up
Calculation:: As published (Table B8, England and Wales total).
Compare with:: 2018–19 peak (5,192)
Not included / breaks:: Counts admission episodes, not people. Injuries treated in A&E without admission are not included. Affected by NHS coding practice, but not by police recording.

Source: ONS, Crime in England and Wales: Appendix tables, year ending March 2026, Table B8 (released 23 July 2026). Chart CRIME-10.

View the data as a table

Year	Admissions
2012–13	3,960
2013–14	3,794
2014–15	3,684
2015–16	4,151
2016–17	4,466
2017–18	5,103
2018–19	5,192
2019–20	4,769
2020–21	4,112
2021–22	4,185
2022–23	3,789
2023–24	3,900
2024–25	3,523
2025–26 (provisional)	3,337

Violence on the Crime Survey

Has violence fallen on the survey?

Survey violence is down from 4.5m incidents in 1995 to 0.99m

Crime Survey incidents of violence with or without injury against adults, millions, 1981 to 2025–26.

England and Wales · 1981 to 2025–26

ONS describes the survey as the best guide to trends in violence for the household population, and it shows a long fall — not an exception to falling crime. Knife offences and homicide are different, rarer measures recorded by the police. Violence is “no statistically significant change” on the previous year.

England and Wales · 1981 to 2025–26 · Accredited official statistics

Sources, definitions and data

Geography:: England and Wales — adults aged 16+ living in households
Measure:: Million incidents
Period:: 1981 to 2025–26
Status:: Accredited official statistics
Calculation:: As published (Table A1a).
Compare with:: 1995 peak
Not included / breaks:: Under-counts domestic violence (victims may not disclose it to an interviewer) and excludes children and homicide. Too few serious incidents to separate out knife violence.

Source: ONS, Crime in England and Wales: Appendix tables, year ending March 2026, Table A1a (released 23 July 2026). Chart CRIME-11.

View the data as a table

Year	Incidents (m)
1981	2.063
1983	2.014
1987	2.265
1991	2.900
1993	3.768
1995	4.464
1997	3.746
1999	3.375
2001–02	2.551
2002–03	2.579
2003–04	2.355
2004–05	2.191
2005–06	2.221
2006–07	2.287
2007–08	1.975
2008–09	1.889
2009–10	1.841
2010–11	2.126
2011–12	1.956
2012–13	1.936
2013–14	1.562
2014–15	1.726
2015–16	1.468
2016–17	1.432
2017–18	1.426
2018–19	1.344
2019–20	1.239
2022–23	0.998
2023–24	1.125
2024–25	1.103
2025–26	0.989

Reading the four measures together

Each measure has a blind spot. The survey covers most violence against adults but too few serious incidents to track stabbings. Police counts cover serious offences but move with recording. Hospital admissions are independent of the police but count only the most serious injuries. Homicide is counted almost completely but is too rare to show short-term change reliably. When all four point the same way — as they do for the latest year — the conclusion is stronger than any one alone.

Violence is not an exception to the long-run fall in survey crime: on the survey it has fallen by about 78% since 1995. The question that remains open is the recent path of the most serious violence, where the recorded series cross method changes.

Evaluated prevention — Violence Reduction Units and hot spot policing — is compared on the crime overview: targeted serious-violence prevention.

Sources and methods (3)

ONS, Crime in England and Wales: Appendix tables, year ending March 2026, Worksheet B6; Tables A5a, B8 and A1a (released 23 July 2026) — police-recorded knife or sharp instrument offences (B6); homicide (A5a); hospital admissions for sharp-object assault (B8); Crime Survey violence (A1a)
ONS, Crime in England and Wales: year ending March 2026 (released 23 July 2026) — knife homicides, force shares, recording notes
Home Office, Violence Reduction Units year ending March 2025 evaluation report (released 2 April 2026) — evaluation of serious-violence funding

Method notes

Knife offences: selected violent and sexual offences in which a knife or sharp instrument was used to injure, used as a threat, or believed present. Possession offences are counted separately and are not included.
NDQIS (National Data Quality Improvement Service) is an automated tool forces use to flag knife involvement; 41 forces used it by March 2026.
Homicide counts are from police-recorded crime and differ from the Home Office Homicide Index, which reflects court outcomes.