Background: Problems such as redundant publication, plagiarism and data fabrication can reduce the reliability of the medical evidence base. Although guidelines are available, little is known about journal editors’ views about such problems or their frequency. Objectives: To determine journal editors’ levels of concern about publication ethics issues and their occurrence at their journals. Methods: A survey (which could be returned by mail or electronically) was sent in June 2007 to 524 editors-in-chief of science journals published by Wiley-Blackwell. The questionnaire asked editors about the severity and frequency of 16 ethical issues at their journals, their confidence in handling such issues, and their awareness and use of published guidelines and other resources. Results: Responses were obtained from 231 editors (44%), of whom 48% edited healthcare journals and 49% other science journals based mainly in Europe (50%) and North America (36%). Half the titles (46%) were published on behalf of academic societies. The general level of concern about the 16 issues was low, with mean severity scores of < 1 (on a scale of 0-3) for all but one. The issue of greatest concern (mean score 1.19) was redundant publication. Most editors felt confident in handling the issues, with < 15% feeling ‘not at all confident’ for all but one of the issues (gift authorship, 22% not confident). Most editors believed the stated problems occurred less than once a year and > 20% of the editors stated that 12 of the 16 issues never occurred at their journal, however 13% to 47% did not know the frequency of individual problems or did not answer these questions. Awareness and use of guidelines and other resources was generally low. Most editors were unaware of all except other journals’ instructions (only 32% had used the most widely known guideline: the Blackwell Best Practice Guidelines on Publication Ethics). Conclusions: Most editors of science journals believe that ethical problems occur only rarely in their journals. However, a considerable proportion admit that they do not know how often such problems occur. Although some would welcome more guidance or training, most editors are unfamiliar with available guidelines.