JOURNAL editors and publishers are not doing enough to limit misconduct and fraud, according to a survey published on Tuesday by COPE, the UK's Committee on Publication Ethics.

Just 118 of COPE's 346 member journals replied to the survey. Two-thirds had no anti-fraud safeguards, 60 per cent had no complaints procedure, and half did not publish author guidelines.

Fraudulent research can slip through even if journals are alert to misconduct. "If the fraud is clever enough, it is likely that referees and editors will not notice it," says Linda Miller, the US executive editor of Nature.

Harvey Marcovitch, chair of COPE, says the committee has seen an increase in enquiries about misconduct since Woo Suk Hwang's cloning fraud was brought to light last year, although there is no evidence that fraud itself is becoming more common. The situation is similar in the US, according to the Office of Research Integrity in Rockville, Maryland.

Marcovitch hopes that the launch this month of the UK Panel for Health and Biomedical Research Integrity in London will help universities and the National Health Service spot misconduct before research is published.