PURPOSE: To investigate the perceptions of scientists and institutional representatives (IRs) to the National Institutes of Health's Office of Research Integrity concerning appropriate punishment for unethical research behavior. METHOD: In 1994-95, 606 scientists and 91 IRs rated the ethical behaviors of and suggested appropriate punishments for protagonists in randomly generated scenarios describing scientific research behaviors. The authors evaluated the relationships of the suggested punishments to the protagonists' behaviors and characteristics, and compared recommendations of the scientists and IRs. RESULTS: The respondents suggested punishments for 80% of the scenarios that were rated unethical. Punishments were more often prescribed for behaviors rated more unethical and for repeat offenders. The type of punishment was related to the protagonist's academic status and the nature of the unethical behavior. IRs proposed more and different punishments than did scientists. CONCLUSION: Scientists and IRs proposed that most unethical research behaviors be punished. The decision to punish depended on the unethical level of the behavior. The type of punishment depended on the aims: correcting the wrong, rehabilitation, or sanction. Variation in the respondents' selections of punishments and the IRs' greater propensity to punish suggest that scientists committing similar ethical violations may receive different punishments. Explicit consideration of which punishment is merited under what circumstances should be undertaken by the scientific community.