Energy analysis software is an essential component of efforts to foster increased energy efficiency in buildings. In North America alone, there exist hundreds of web- and disk-based building energy analysis tools, serving a diversity of audiences. Some are specialized while others consider the building as a whole. We evaluated 50 web-based residential tools and 15 disk-based tools. While the state-of-the art in tool design has risen considerably over the past three decades, today's users are faced with an increasing-and often overwhelming-array of choices and, often, conflicting results. A surprising number provide little or no detailed analysis of energy savings options. A number of important building energy issues and efficiency features cannot be sufficiently well evaluated using any of the existing tools. Many factors conspire to confound performance comparisons among tools, and the sources or implications of observed differences in results are difficult to pinpoint. For the tools we tested, predicted whole-house energy bills ranged widely (by nearly a factor of three), and far more so at the end-use level. We also discovered a remarkable number of indications of errors in programming or algorithms. Tool design should be grounded in social science and engineering. Analytical results (e.g., benchmarking) and end-use-specific "what-if" functions are more helpful for many users than ratified engineering outputs. Desirable technical features include modeling of occupant effects, open-ended energy calculations as well as results normalized to actual consumption history, incorporating means for users to grasp the uncertainties embodied in the results, and ensuring quality control to remove errors from the design and programming of tools. More coordinated planning of tool development could help address the fragmentation and dilution of efforts that has historically hampered tool quality and market penetration. (C) 2004 Elsevier B.V. All rights reserved.