The ideal of scientific openness — i.e. the idea that scientific information should be freely accessible to interested parties — is strongly supported throughout the scientific community. At the same time, however, this ideal does not appear to be absolute in the everyday practice of science. In order to get the credit for new scientific advances, scientists often keep information to themselves. Also, it is common practice to withhold information obtained in commissioned research when the scientist has agreed with his (...) or her employer to do so. The secrecy may be intended for ever, as in the military area, but also temporarily until a patent application has been made. The paper explores to what extent such secrecy is undesirable, as seems to be suggested by the ideal of scientific openness. Should this ideal be interpreted as having certain exceptions which make the above-mentioned practices acceptable? Are there, on closer inspection, good arguments for the ideal of scientific openness, and for officially upholding it? Possible versions of the ideal of scientific openness are explored and the issue is found to be rather complex, allowing for wide variations depending on the acting parties, beneficiaries, types of information and moral requirements involved. We conclude that the arguments usually given in favour of this ideal are weaker than what seems to be generally believed, and that, on closer inspection, they leave plenty of room for exceptions to it. These exceptions only partly cover the actual practice of withholding scientific information, and there may still be good reason to advocate, teach and enforce the ideal of scientific openness within the scientific community. (shrink)