The Multilevel Evaluation Systems project was field tested in grades 4 and 5 classrooms in five districts of the University of Pennsylvania's School Council. Multilevel evaluation coordinates multiple sources of information (about student characteristics, school context, school and instructional processes, and a range of student outcomes) with multiple reporting targets (students, classes, schools, school boards, districts, state and federal government) to facilitate data-based decision-making at all levels. The field study attempted to set up user-based data collection by first identifying needs and concerns common across groups as well as concerns unique to each district. The core database included variables addressing: (1) background information about students; (2) information on student outcomes, both academic and affective; (3) classroom processes; and (4) school content including parent involvement. The types of reports desired by district superintendents, school principals, and teachers were compared. District leaders and principals favored analysis of trends over time, whereas teachers wanted a detailed breakdown of student characteristics and current achievement. All groups preferred that a one-page synthesis of data be designed to meet their particular needs. Steps for insuring quality control in data collection and for providing support in data interpretation and use were recommended. (LPG)