Evaluation units found in large organizations confront a number of impediments to the use of their findings and recommendations. These derive from both the formal roles they play and their own operating styles and processes. Drawing on experience at the U.S. General Accounting Office and the World Bank, this paper suggests that to improve the usefulness--and use--of their work, such units need to pay careful attention to the multiplicity of audiences they necessarily address, the divergent information needs of those audiences, and the timeliness of their work.