The last 15 years have seen an explosive growth in university-based research activities in the fields of computer research and information theory. Technology has both driven and been driven by these activities. Development of the technology for very large-scale integration (VLSI), for example, has driven research on special purpose processors and highly parallel computers which has been translated into actual products that have expanded the technology base and have driven the research further. As a consequence of the push/pull of technology and basic research, many researchers in computer science, computer engineering, and information science find themselves working on basic research problems that may have short-term applications important to the commercial sector and/or important to the national defense. These researchers may face the problem of separating their research into two components: aspects dealing with basic research which should be fully available to their colleagues, and aspects with immediate applications which, for reasons of commercial importance or national defense, should not be made fully available. This problem of separation could lead to problems in the exchange of scientific information by the researchers. External controls on research, either from government or industrial sponsors, could also affect the exchange of scientific information. To study the scientific communication patterns influenced by a sponsor's actual or potential controls, I undertook a case study of three research groups in computer science. The groups were chosen to be illustrative of research in which secrecy could play a restrictive role in scientific communications. The three groups-in cryptology, systems software, and VLSI research-are located at three different U.S. universities, all of which are major research universities with strong research groups in computer science. I interviewed the principals of all three research groups to determine the existence of any controls and the effect of those controls on the conduct of scientific research. Questions were also posed to determine how development work was separated from basic research. Basic research in computer science may be subject to a number of controls, both governmental and industrial. In the United States, governmental controls consist of the Invention Secrecy Act' (which applies to the patenting of militarily useful products), the International Traffic in Arms Regulations2 (which apply to military products), the Export Administration Regulations' (which apply to products and technology that are both militarily useful and useful for the civilian purposes), and the President's 1982 Executive Order 123564 that defines classification for national security purposes. Some research contracts also contain pre-publication review conditions. Industrial controls consist of contractually agreedupon conditions, such as exclusive licenses or pre-publication review, and publication delays to ensure the patentability of products. The latter is also written into many university employment contracts. Conflicts of interest that also affect scientific John C. Cherniavsky is Professor of Computer Science, Computer Science Department, Georgetown University, Washington, DC 20057. This essay was produced with the support of funds from the National Science Foundation and the National Endowment for the Humanities under NSF Grant RII83098 74.