Crime & safety · Trends · England and Wales · 1981 to 2025–26

What has fallen, and by how much?

On the Crime Survey, crime excluding fraud and computer misuse fell from 19.8m incidents in 1995 to 4.33m in the year to March 2026 — about 78% lower. Theft, criminal damage and violence each fell by more than 70% from their peaks. Police-recorded crime moved differently, because recording changed.

Evidence last reviewed 3 October 2026

The comparable survey series

Which crimes fell?

Theft, criminal damage and violence are all down by more than 70% from their peaks

Crime Survey for England and Wales incidents by main crime type, millions, 1981 to 2025–26.

England and Wales · 1981 to 2025–26

Theft peaked at 11.6m in 1995 and is now 2.6m; violence fell from 4.5m to 0.99m. The fall was steepest from the mid-1990s to the mid-2010s and has been much flatter since: the July 2026 ONS bulletin reported no statistically significant annual change in theft or violence for the year ending March 2026. For the crimes it covers, the survey is a better trend guide because it is not affected by changes in reporting or recording.

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

Sources, definitions and data

Geography:: England and Wales — adults 16+ (violence, robbery, personal theft) and households (burglary, household and vehicle theft, criminal damage)
Measure:: Million incidents
Period:: 1981 to 2025–26
Status:: Accredited official statistics
Calculation:: As published (Table A1a, thousands, shown in millions). The four types are parts of the same survey total, measured the same way, so they share one axis.
Compare with:: Each type’s own peak
Not included / breaks:: No estimates for 2020–21 and 2021–22. Survey years to 1999 are calendar years; from 2001–02 they are interview years ending March. Fraud and computer misuse are not in these series.

Source: ONS, Crime in England and Wales: Appendix tables, year ending March 2026, Table A1a (released 23 July 2026). Chart CRIME-04.

View the data as a table

Year	Theft (m)	Criminal damage (m)	Violence (m)	Robbery (m)
1981	6.29	2.78	2.06	0.164
1983	7.14	2.83	2.01	0.145
1987	8.23	2.98	2.27	0.178
1991	9.79	2.68	2.90	0.195
1993	11.51	3.30	3.77	0.254
1995	11.62	3.39	4.46	0.322
1997	10.02	3.06	3.75	0.338
1999	8.58	2.90	3.38	0.394
2001–02	7.21	2.67	2.55	0.340
2002–03	6.93	2.59	2.58	0.290
2003–04	6.52	2.47	2.35	0.269
2004–05	5.82	2.53	2.19	0.245
2005–06	5.66	2.68	2.22	0.301
2006–07	5.64	2.94	2.29	0.318
2007–08	5.13	2.63	1.98	0.304
2008–09	5.58	2.70	1.89	0.264
2009–10	4.99	2.40	1.84	0.319
2010–11	5.17	2.17	2.13	0.239
2011–12	5.36	2.03	1.96	0.244
2012–13	4.88	1.74	1.94	0.199
2013–14	4.38	1.45	1.56	0.166
2014–15	4.03	1.34	1.73	0.088
2015–16	3.70	1.21	1.47	0.153
2016–17	3.33	1.17	1.43	0.129
2017–18	3.58	1.08	1.43	0.160
2018–19	3.75	1.15	1.34	0.192
2019–20	3.30	1.00	1.24	0.127
2022–23	2.65	0.66	1.00	0.072
2023–24	2.69	0.66	1.13	0.113
2024–25	2.80	0.61	1.10	0.083
2025–26	2.63	0.65	0.99	0.071

Scroll the table sideways to see every column.

The Crime Survey asks a large random sample of adults what happened to them in the past year, whether or not they told the police. Its methods have stayed comparable since 1981, so it is the measure to use for “has crime gone up or down?” — for the crimes it covers. On that measure the big household crimes of the 1990s, burglary, vehicle crime and criminal damage, fell furthest. Violence fell by about 78% from its 1995 peak.

Why crime fell is debated, and the survey itself does not test causes — so this page does not attribute the fall to any policy. What is clear is that it has slowed: ONS finds no statistically significant change in theft, violence or the headline total on the previous year.

Two things sit outside these lines. Fraud and computer misuse were added to the survey only in 2016–17, and are now 55% of the headline total — see fraud and online harm. And rare, serious crimes such as homicide and knife violence are too infrequent to measure in a household survey — see serious violence.

Why police figures moved the other way

Why do police figures tell a different story?

Police-recorded crime rose from 3.5m to 5.2m as recording improved

All offences recorded by the police in England and Wales excluding fraud and computer misuse, millions, 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.

Recorded crime fell until 2013–14 and then rose while the survey kept falling. ONS says the large rises in recorded violence since 2015 were mainly because of improvements to recording practices, not more crime. It is a different measure from the survey, covering more offence types and populations, so the two lines should not be compared level for level — this is why it is shown on its own axis.

England and Wales · 2002–03 to 2025–26 · Official statistics (not accredited)

Sources, definitions and data

Geography:: England and Wales — All recorded offences: individuals of any age, households, businesses and society
Measure:: Million offences
Period:: 2002–03 to 2025–26
Status:: Official statistics (not accredited)
Calculation:: As published (Table A5a).
Compare with:: Its own low point, 2013–14
Not included / breaks:: Recording practice tightened substantially from 2014 after inspections found under-recording, especially of violence. Crimes not reported to the police are not included.

Source: ONS, Crime in England and Wales: Appendix tables, year ending March 2026, Table A5a (released 23 July 2026). Chart CRIME-05.

View the data as a table

Year	Recorded offences (m)
2002–03	5.791
2003–04	5.844
2004–05	5.477
2005–06	5.426
2006–07	5.322
2007–08	4.881
2008–09	4.630
2009–10	4.265
2010–11	4.078
2011–12	3.904
2012–13	3.553
2013–14	3.507
2014–15	3.575
2015–16	3.899
2016–17	4.322
2017–18	4.890
2018–19	5.266
2019–20	5.305
2020–21	4.616
2021–22	5.327
2022–23	5.578
2023–24	5.403
2024–25	5.307
2025–26	5.235

Only about 36% of the incidents the survey picked up in the year to March 2026 (excluding fraud and computer misuse) were reported to the police. Recorded crime therefore depends on what people report and how forces record it. After inspections in 2014 found crimes going unrecorded, forces tightened their practice, and recorded violence and public-order offences rose without any matching rise in the survey.

That does not make recorded crime useless. It covers crimes against businesses and children, and rare, serious offences the survey cannot measure, and it is the only source for local figures — which is what the area explorer uses. It just cannot be read as a trend in how much crime there is.

What would change the picture is compared on the crime overview — What would move the needle?

Sources and methods (2)

ONS, Crime in England and Wales: Appendix tables, year ending March 2026, Tables A1a and A5a (released 23 July 2026) — Crime Survey incidents by type (A1a); police-recorded crime (A5a)
ONS, Crime in England and Wales: year ending March 2026 (released 23 July 2026) — significance of changes, reporting rate, recording caveats

Method notes

Crime Survey years to 1999 are calendar years; from 2001–02 they are interview years ending March, covering crimes in the 12 months before interview.
No survey estimates exist for 2020–21 and 2021–22: face-to-face interviewing was suspended during Covid. The lines break there.
Survey figures are not comparable with those published before January 2019, when ONS changed how repeat victimisation is counted; this page uses the current back series.