Murderers could be escaping justice because of blunders at the scenes of deaths, an official report suggests.
Police and non-forensic pathologists attending unexplained fatalities were said to have overlooked "obvious indicators of suspicion".
Bodies did not appear to have been inspected at all in five instances.
The suspicious cases in a sample of 32 were ultimately picked up but a Home Office study published today said it "seems reasonable to suspect that homicide cases may have been missed in the past, and could continue to be missed or forensic evidence lost".
An additional 150 cases reported since January 2014 should now be examined, watchdogs said.
The extraordinary warnings came in a study ordered in the wake of concerns raised by a routine audit in 2012 of cases in which non-forensic post-mortem examinations were started before being stopped in favour of a forensic procedure.
Initial assessments indicated that all should have been seen as requiring more rigorous forensic autopsies from the outset, prompting the forensic science regulator to ask the Home Office's Forensic Pathology Unit to conduct further investigations.
Of the 32 cases, ten transpired to be homicides and a further five were suspicious deaths requiring further investigations.
The report said: "Of the 10 confirmed homicides the failure to identify them as such was as a consequence of initial police decision making in seven cases.
"Of the five suspicious deaths requiring further investigation, failure to identify them as such was as a consequence of police decision making in 2 cases."
Overall, decisions of police were said to be "questionable" in 15 - or 47% - of all cases examined.
There were nine episodes where police were not initially involved in the investigation and did not attend the scene.
In ten there was no reason to consider the death suspicious from the information at the scene, including three which were later confirmed as homicide and one suspicious fatality.
The report said: "This highlights how difficult it can be at some scenes in terms of the scene assessment and decision-making process."
Authors said it was suspected that "cognitive bias" may have influenced decisions.
One unnamed force visited by the research team identified that alcohol and drugs were a feature which had "influenced officers to make isolated decisions that the cause of death was not suspicious".
It went on: "Officers appeared to presume that death was as a result of alcohol consumption leading to injury through falling or some other cause due to intoxication."
Another factor highlighted in the study were inspections of bodies at the scenes of death.
There was some evidence that even though there were visible marks indicating possible violence, decisions were made not to treat the deaths as suspicious. In five there "appears to have been no inspection of the body at all", the report said.
Senior officers also suggested that finance is a consideration but the study found no evidence to indicate that cash constraints were a factor in the failure to use a Home Office registered forensic pathologist (HORFP).
Cases examined were said to be examples where the "system worked" and killings, which may otherwise have gone undiscovered, were identified.
However, the report said: "The findings of this limited study of cases ... have highlighted the potential for professionals involved in death investigation to 'miss' homicides by conducting limited scene assessments and not utilising the advice and expertise of a HORFP early on in the investigation process.
"The prevailing situation poses an obvious threat to the criminal justice system due to factors such as inadequate training, cognitive bias and financial pressures."