In addition to the long-standing requirements for accuracy, timeliness, and reliability, there is one more essential requirement for the dissemination of statistical data : The information presented must be intelligible to the intended audience. This poses a significant challenge to all producers of economic statistics. As the audience has grown larger it has also grown much more diverse. Presenting complex data in a form that can meet the differing needs of a highly varied user population is a non-trivial task. What is at stake here are the human factors, or usability, of a particular screen design or sequence of screens. Like other software engineering methodologies, usability engineering includes requirements gathering, design, implementation, and testing phases. Usability testing is the process by which the human-computer interaction characteristics of a system are measured, and weaknesses are identified for correction. Such testing can range from rigorously structured to highly informal, from quite expensive to virtually free, and from time-consuming to quick. The resources required to effectively implement usability engineering into a Web site development effort fall into three main categories: staff, time, and money. There are ways to reduce the required expenditures in each of these areas, and it is quite feasible to start small and add resources over time as the concrete benefits of usability testing emerge through use.