The scientific establishment is deeply concerned over a proposed regulation that would require data to be shared on projects that are federally funded. Specifically, the proposed amendment to OMB Circular A-110 would require data collected by researchers at universities, hospitals, and non-profit institutions to be shared with interested parties if (1) the data are produced as part of a grant or agreement funded by the federal government; (2) the data are used in a published study; and (3) the data or study is used in formulating a policy or rule. Parties could request the data under the Freedom of Information Act. The proposed rule responds to a provision by Senator Richard Shelby in the 1999 Omnibus Spending Bill that requires data generated under federal awards at universities and non-profit institutions to be available to the public. This regulatory analysis develops an economic framework for evaluating proposals to provide greater access to research data. Our analysis also offers specific recommendations for improving OMB Circular A-110 as well as the broader regulatory process. We argue that the economic analysis of sharing research findings can be separated into three parts: the impact of requiring public access on incentives to produce data, research, and innovation; the impact of that requirement on the quality of research; and the impact of required access on the efficiency and transparency of policy. The economic analysis demonstrates that the standard property-rights framework used to justify time-limited property rights for the use of data is not sufficient for addressing broader problems in which research and data could be used to help inform public policy decisions. The value of sharing data for public policy must also be considered. A second conclusion is that traditional peer review done by scientific journals is not adequate for purposes of relying on research for major public policy decisions. A third conclusion is that scientists who are reluctant to share their findings are more likely to have errors in their analysis than the average researcher. A fourth conclusion is that requiring the release of data could slow the development of data and delay the publication of results. Although substantial costs and uncertainty may be associated with greater public access to data, our analysis suggests that academic norms alone provide very limited access to scientific data. We recommend improving Circular A-110 by narrowing and clarifying the scope of the proposed regulation. The proposed regulation should apply to economically significant regulations that have an annual economic impact of at least $100 million. In addition, we recommend that Congress create an agency that would be charged with replicating the findings of regulatory agencies before such regulations could be implemented. The recommendations concerning replication would require additional legal authority. Taken together, our recommendations would help lay the foundation for a regulatory system that is more accountable and has more scientific integrity.