Britain's academy of science is to set up an inquiry into how scientific research is made public.
It follows rows about the reliability of some studies which, although they were published in journals, were later found to have been based on false or poorly interpreted results.
Editorial consultants resigned from an online journal when it published cloning claims
There is also concern about organisations which make scientific claims in press releases and at media conferences but then present no evidence to support their announcements.
A working party for the Royal Society will look at how research information that could influence public opinion and policy is checked.
It may then recommend changes and some "best practice" guidelines for scientists.
Most scientific research is "peer reviewed" with other experts analysing the results before they are published in a scientific journal.
The process is supposed to ensure that any study's methodology is sound and that interpretation of data does not go beyond what can be reasonably justified.
Peer review is intended to be a "gold standard" that protects other scientists and the public from shoddy research and fraudulent claims.
But the system does not satisfy everyone, and there is some concern that a number of journals may be publishing research just to grab headlines in the mainstream media.
"I worry it has gone to tabloid newspaper-like battles about silly things rather than focussing on the deeper issues of the science and the real benefits of what it all means," Professor Robin Lovell-Badge told the BBC.
The stem cell researcher resigned from the editorial board of one online journal after it published claims on human embryo clones that "had no scientific value".
The Royal Society's vice president Sir Patrick Bateson, who will chair the working group, added: "Peer review has been criticised for being too secretive, conducted behind closed doors and assessed by anonymous referees".
He said it had also been suggested the process had been used by the establishment to "prevent unorthodox ideas, methods and views, regardless of their merit, from being made public".
Green groups are not convinced about the impartiality of government science
But Professor Bateson stressed that peer review was a good form of insurance, and anyone who encountered a study - particularly journalists - should always ask if the science had been checked by others.
"Stories that get into the media that haven't been properly reviewed can do enormous damage," he said.
"The MMR triple vaccine was a particularly bad case where people were terrified because they thought that vaccinating their children would lead to autism, and the evidence for that was terrible."
Row over GM crops
The Royal Society panel will focus particularly on the reliability of checks on research funded by the government, large companies and pressure groups.
There are concerns that conflicts of interest may influence how a study is written up - the interpretation put on data could be slanted.
Anti-GM campaigners, for example, are deeply sceptical about the science commissioned by the UK Government to assess the impact of novel crops.
They believe the government is determined to push ahead with GM farming whether the public wants it or not and that the science will inevitably turn out to be supportive.
Sir Patrick said scientists used peer review "almost exclusively" to publicise findings. But he said researchers could still attract publicity "for highly questionable results even when they offered no evidence that their research had been checked".
This was evident earlier this year when the Raelian sect announced the births of human clones.
The only proof the sect's US-based company Clonaid produced to support its assertion was a photograph of one of the children alleged to have been born in Japan.
Many scientists dismissed this and demanded DNA evidence. None was provided.
The Raelians "published by press conference"
"If you publish by press conference, you are bound to allow bad research to get out into the media," Professor Bateson said.
The Royal Society's working group on scientific research includes leading journal editors, such as Dr Phillip Campbell from Nature and Dr Andrew Sugden from Science, as well as journalists and academics working in the field of science communication.
Submissions can be made to the panel up to 26 September.
A "best practice" guide for scientists may then be devised.