Computer-assisted testing is not without its problems and pitfalls, but it holds a great deal of promise as well. Computer administration of tests provides more control over the testing process than was ever possible with paper-and-pencil testing. At the same time it offers the possibility of being able to monitor and record aspects of the testing process, such as response latency and response shifting, that may prove to be important predictive factors in their own right. Computer scoring of tests has made it possible to obtain accurate scores. It has been estimated that errors involving a difference of one or more points in the final score are made in 10 percent of cases involving hand scoring of objective tests. These and similar errors of measurement may have more impact on the reliability of scores obtained in practice than some of the better analyzed sources described in measurement theory. In the final analysis, computer interpretations of test scores may offer the greatest potential for advancing psychological measurement. As the volume of research data relevant to a particular test increases, the task of using it effectively in interpretation becomes increasingly frustrating for the unassisted test user. Perhaps even more importantly, computerized reports produce consistent, predictable outputs that can be analyzed and improved if the appropriate models and techniques are developed for doing so and they are treated scientifically and not as scientific curiosities. (ABL)