The incidence of homicide by strangers is regarded by experts as one of the most reliable measures of the true rate of violence in society because the figures are difficult to manipulate.

The disturbing statistics revealed today suggest a rise in random incidents of violence rather than a growing weapons culture of blades or firearms. Stranger killings have rocketed in the West Country and London but dropped sharply in Greater Manchester, where police strive to identify and monitor the most dangerous individuals.

A close examination by the Crown Prosecution Service of stranger killings in one of the harder-hit areas, Devon & Cornwall, has found that people are as likely to be killed by a stranger's bare hands or boots as knives or guns. Yet across the UK the number of people who die at the hands of somebody they know has remained remarkably constant at about 400 a year.

Of those, crimes of passion stayed steady at around 130 annually. In a surprising development, this means people are now as likely to be killed by a complete stranger as by their husband, wife, lover or a former partner.

Other killings by people who know their victim include 30 a year by parents on children and 22 by children on parents. Most “known” killers — 213 a year — are friends, acquaintances or relatives beyond the nuclear family.

The increase in stranger homicide chimes with research by Manuel Eisner, deputy director of the Institute of Criminology at the University of Cambridge, who has found that young men are increasingly being killed by strangers. His findings have been linked to young male group behaviour and leisure patterns.

Dr Eisner told The Times that the increase “is one of the best available indicators of serious violence in public space.”

Officials define a stranger killing as one where “the principal subject is not known to the victim”.

The figures do not include the 52 people who were killed in the July 7 attacks by suicide bombers in 2005 nor the victims of the mass murderer, Dr Harold Shipman.

Richard Garside, director of the Centre for Crime and Justice Studies at King’s College, London, said: “We have seen an increase in homicide since the mid-1990s and I think it is a reasonable argument that this shows there is a lot of hidden violence which is potentially going on but is not being picked up in the statistics.”

He added: “What we know about the long-term trends in homicide is that the risk of being a victim of homicide in the bottom 10 per cent of society has gone up an awful lot and the risk at the top end has fallen.

“There is a lot of violence about that is not officially recognised by government statistics. For every homicide there will have been many violent incidents that may have led to homicide if it hadn’t been for intervening factors,” he said.

The geographical pattern of stranger killings reinforces the findings of criminologists that gun crime is unlikely to be responsible for the rise.

Greater Manchester has seen a rapid fall in stranger killings, attributed to a reduction in firearms crime, possibly due to the introduction of a five-year minimum sentence for possession. Nottinghamshire, notorious nationally for gun crime, has a low overall toll of stranger killings.

Only 99 people were killed by strangers in the year Labour took office. Figures show a steadily rising trend reaching 130 such homicides last year. The total body count of stranger killings in Labour’s first eight years is 903.

Devon & Cornwall has seen the largest rise. While there was only one stranger killing, or none, a year in the late 1990s, there have been 26 in the past five years. Sharp but smaller rises can also be seen in Dorset, Avon & Somerset and Gloucestershire.

Greater London saw a doubling from 57 in Labour’s first four years to 101 in the next four. Greater Manchester’s toll has tumbled from a peak of 37 in 1999-2000, to only 5 and 7 stranger killings in the past two years respectively.

Nottinghamshire is relatively safer. The force suffers few stranger killings, peaking at 6 in 1999-2000 and falling to just one last year.