Procedures for generating evaluation questions in the Discrepancy Evaluation Model (DEM) are problematic as they are described in the literature. These procedures would lead to unmanageably or ineffectively long lists of questions. Any process of narrowing down the list must be drawn from other sources. One such source, developed for a specific project, focuses on maximizing utilization potential. This procedure essentially takes the outcomes of the program's Input-Process-Output charts to determine the likelihood that information about each outcome will be useful for decisions or policy considerations. Key program staff answer each question with a scaled response. The responses are tallied and used to stimulate discussion among the staff. Consensus is sought on priority outcomes and concerns around each from which the evaluators develop the evaluation questions. The paper presents this procedure in more detail, an example of its use, and discusses the benefits from its application to a major employment training technical assistance program. (PN)