Objectives. - To identify the prevalence and determinants of data- withholding behaviors among academic life scientists. Design. - Mailed survey of 3394 life science faculty in the 50 universities that received the most funding from the National Institutes of Health in 1993. Participants. - A total of 2167 faculty responded to the survey, a 64% response rate. Outcome Measures. - Whether respondents delayed publication of their research results for more than 6 months and whether respondents refused to share research results with other university scientists in the last 3 years. Results. - A total of 410 respondents (19.8%) reported that publication of their research results had been delayed by more than 6 months at least once in the last 3 years to allow for patent application, to protect their scientific lead, to slow the dissemination of undesired results, to allow time to negotiate a patent, or to resolve disputes over the ownership of intellectual property. Also, 181 respondents (8.9%) reported refusing to share research results with other university scientists in the last 3 years. In multivariate analysis, participation in an academic-industry research relationship and engagement in the commercialization of university research were significantly associated with delays in publication. Odds ratios (ORs) and 95% confidence intervals (CIs) were 1.34 (1.07-1.59) and 3.15 (2.88-3.41), respectively. Variables associated with refusing to share results were conducting research similar to the Human Genome Project (OR, 2.09; 95% CI, 1.75-2.42), publication rate (OR, 1.02; 95% CI, 1.01-1.03), and engagement in commercialization of research (OR, 2.45; 95% CI, 2.08-2.82). Conclusions. - Withholding of research results is not a widespread phenomenon among life-science researchers. However, withholding is more common among the most productive and entrepreneurial faculty. These results also suggest that data withholding has affected a significant number of life-science faculty and further study on data- withholding practices is suggested.