CCTV schemes in town centres do not stop drunken street violence breaking out, according to new research.
The study was the first to compare police and hospital data
But cameras do alert police to assaults and reduce the number of people treated at casualty departments.
Scientists also say CCTV has reduced the severity of injuries suffered in street brawls.
But the study, published in the Injury Prevention journal, concludes there is no evidence of the surveillance systems having a deterrent effect.
It says: "The benefit of CCTV might lie less in preventing such offences... but more
in facilitating a faster police response to arguments or assaults in public
spaces, which limits their duration and therefore reduces the incidence and
seriousness of injury."
Experts from the University Hospital of Cardiff, who carried out the research, also concluded official police statistics on violent crime were inadequate and "inappropriate".
They found police statistics recorded only a quarter of assaults leading to treatment in casualty departments.
The authors of the study say it was the first to compare police and hospital data and that its four year time span was longer than other CCTV evaluations.
They studied police reports of street violence from 1995 to 1999 in five randomly chosen towns where CCTV was installed in 1997 - Ashford, Eastbourne, Lincoln, Newport in the Isle of Wight and Peterborough.
The data was compared with towns that had no surveillance cameras at the time - Chelmsford, Poole, Derby, Scarborough and Huntingdon.
They then checked casualty department records for the treatments of assaults over the same period.
In the areas with CCTV the number of people treated for injuries after assaults fell by 3%, while the number of violent offences detected by police rose by 11%.
In the towns not covered by CCTV the numbers needing treatment rose 11%, but violent offences detected by police remained the same.
The research paper concluded that, if CCTV surveillance did act as a deterrent, police detection rates would have fallen.
Co-author Jonathan Shepherd, a professor of oral and maxillofacial surgery, said: "The evidence shows you can't rely on police violence statistics as an accurate measure of violence in the community.
"True measures have got to take into account injury data from local hospitals as well as police information."
The findings echo government research published last year that concluded CCTV was not as useful in the fight against crime as was previously thought.
That research also concluded that better illumination could be a cheap way of cutting illegal activity, especially in crime hotspots.