"Police records, rather than representing a reliable measure of trends in violence, are a product of police activity," write Professor Jonathan Shepherd and Dr Vaseekeran Sivarajasingam, from the Violence Research Group at Cardiff University.
According to the police, higher rates of crime are attributable to better surveillance, including CCTV and higher numbers of police in city centres, and changes in recording practices, say the authors.
But in the US, where convictions for violent crime have risen as a result of increased police activity, the actual episodes of violent crime have fallen, they say.
And figures from the British Crime Survey show that violent crime has fallen 36% since its peak in 1995 in Britain.
Although the crime survey has its own limitations, say the authors, its reporting has been consistent. And its findings are backed up by information from hospital emergency departments in England and Wales.
This shows that the number of injuries sustained through violence that was not self inflicted fell 12% in the five years to 2004.
Despite the rising trends in violence reported by the police, they do not capture large numbers of violent incidents, say the authors.
They cite previous UK and US research, showing that violence resulting in injuries, even firearm injuries treated in emergency care, is often not recorded by the police.
"This mismatch between the only local criminal justice source of data (police statistics) and injury data has major implications for police and practice," they write.
"[Police records] should not be used as a measure of underlying trends in violence," they write. "Far from representing a cause of concern, increases in violence recorded by the police are often associated with less harm." Emergency care data are much more reliable and objective, they say.