Biological scientists and reviewers of research integrity have similar attitudes about research norms and professional ethics. Researchers surveyed 606 scientists receiving federal research grants and 91 institutional representatives (IRs) to the federal Office of Research Integrity about scenarios of questionable research conduct. Most thought that the creation or falsification of data and plagiarism were very serious unethical actions. IRs proposed more punishment for ethical violations than the scientists. Respondents felt first offenses were less serious than repeat offenses., Context. -- The professional integrity of scientists is important to society as a whole and particularly to disciplines such as medicine that depend heavily on scientific advances for their progress. Objective. -- To characterize the professional norms of active scientists and compare them with those of individuals with institutional responsibility for the conduct of research. Design. -- A mailed survey consisting of 12 scenarios in 4 domains of research ethics. Respondents were asked whether an act was unethical and, if so, the degree to which they considered it unethical and to select responses and punishments for the act. Participants. -- A total of 924 National Science foundation research grantees in 1993 or 1994 in molecular or cellular biology and 140 representatives from the researchers' institutions to the US Department of Health and Human Services Office of Research Integrity. Main Outcome Measures. -- Percentage of respondents considering an act unethical and the mean malfeasance rating on a scale of 1 to 10. Results. -- A total of 606 research grantees and 91 institutional representatives responded to the survey (response rate of 69% of those who could be contacted). Respondents reported a hierarchy of unethical research behaviors. The mean malfeasance rating was unrelated to the characteristics of the investigator performing the hypothetical act or to its consequences. Fabrication, falsification, and plagiarism received malfeasance ratings higher than 8.6, and virtually all thought they were unethical. Deliberately misleading statements about a paper or failure to give proper attribution received ratings between 7 and 8. Sloppiness, oversights, conflicts of interest, and failure to share were less serious still, receiving malfeasance ratings between 5 and 6. Institutional representatives proposed more and different interventions and punishments than the scientists. Conclusions. -- Surveyed scientists and institutional representatives had strong and similar norms of professional behavior, but differed in their approaches to an unethical act.