From the papers presented at the colloquium, the comments of discussants, and the ideas developed in break-out groups, it was clear that although we have made substantial progress in CBT during the past few decades, much additional work needs to be conducted. Part of this research might be considered evolutionary. A great deal of research is needed comparing various options for the test model, administrative procedure, and scoring so that test developers can make informed judgments on the basis of empirical research.