Scientific misconduct, in recent years, has drawn intense press coverage and substantial policy attention. Various explanations have been advanced to account for its occurrence, ranging from the individual deviance of scientists to the collective transformation of contemporary science, and though each explanation illuminates a facet of the phenomenon, all seem incomplete in some crucial respect and suffer from a dearth of empirical evidence. In this article I briefly review definitions of misconduct and available evidence about its prevalence and critically examine three sorts of explanations: individual psychopathology, anomie, and alienation. Drawing on recent theories of social control, I then suggest a perspective that focuses on the social response to misconduct, not the causes of misconduct itself. I close by outlining some of the implications of this perspective for and policy. Definitions of Scientific Misconduct Are Vague and Unsettled Before a phenomenon can be measured or explained it must first be defined, yet definitions of scientific misconduct are vague, changeable, and disputed [7, 20, 26, 34]. It is difficult to settle on a definition of scientific misconduct because are not well codified or understood, the gradations between proper (even exemplary) technique and misconduct are subtle, and the of settling on a definition is muddled by considerations of the politics and public image of science [10, chap. 5]. Behaviors that scientists may be willing to accept among themselves (concerning, for example, the selection of data for publication or the proper recording and storage of laboratory notes) may be judged unacceptable by laypersons. Even within science, the acceptability of certain behaviors has varied over time, from field to field, and by the state of knowledge within a field [40]. At present the quandary of definition has taken its most pointed form in a dispute between proponents of a quite limited definition of misconduct (as put forth in a recent National Academy of Sciences report [34]) and those who favor a more inclusive definition (as employed by the Public Health Service [12] and the National Science Foundation [7,8]). The Academy report argues that the definition of misconduct in science should be limited to falsification or plagiarism in proposing, performing, or reporting research [34, p. 27]. Questionable practices, including failure to retain good records, unwillingness to disclose or share data, improper allocation of credit, and the misrepresentation of speculations as facts are relegated to a category of lesser severity because such behaviors not directly damage the integrity of the process [34, p. 28]. Harassment, misuse of funds, vandalism, gross negligence, and similar misbehaviors are placed by the Academy panel in a third category because remedies exist for these generic forms of misconduct. For the NSF, by comparison, scientific misconduct includes fabrication, falsification, plagiary and other serious deviation from accepted practices [7, p. 647]. This more comprehensive definition is intended to reflect the ethics of the scientific community and to provide a forum for adjudicating a broad range of complaints that actually arise in the process. For example, of the 124 cases considered by NSF during a three-year period, only 10 had to do with data fabrication or misrepresentation, whereas 70 concerned intellectual property rights ranging from plagiary to theft of ideas [and] failure to give credit [7, p. 584]. The inclusive definition has the advantages of clarity and simplicity -- there is only one type of misconduct, not three. By addressing a greater range of behaviors that palpably harm the enterprise, the inclusive definition may better serve the scientific community's need to air and resolve most complaints on its own, thus reinforcing the professional autonomy and self-regulation of scientists. …