The next World Conference on Research Integrity in Montreal, Canada, on 5–8 May will make collaborators more responsible for the integrity of their research (see go.nature.com/lsd1p5). I believe that more pressure should also be brought to bear on scientific journals, which should publicly declare and reinforce their policies on fraudulent reporting of research results.

Journals were urged in 2010 to improve procedures for tackling allegations of misconduct and irresponsible research practices (www.singaporestatement.org). But progress has been unsatisfactory: some 40% of high-impact biomedical journals, for example, do not have authorship policies, let alone policies to define, prevent and punish misconduct (X. Bosch et al. PLoS ONE 7, e51928; 2012).

There is little excuse for this failure to act against a common, long-standing problem. Editorial associations and publishers have established guidelines on editors' responsibilities regarding suspected or confirmed misconduct in papers (see go.nature.com/egc43n). Automatic detection of plagiarism and image manipulation is now widespread, and compulsory disclosure of financial and non-financial conflicts of interest is becoming standard practice.

Legal disputes and other complications can embroil journals that do not publicly state their policies on misconduct. Worse, those journals serve the scientific community badly.