Most systems of administrative control include both formal and informal, explicit and implicit evaluative mechanisms that sense deviation from standards and activate corrective actions to return subordinates to acceptable levels of performance or to correct beliefs. A review of prior studies showed, however, that few school districts have specific standards against which they assess principal performance and outputs. This paper describes a study aimed at correcting some of these deficiencies (1) by employing the Natriello and Dornbusch conceptual model of evaluation, which identifies clearly delineated evaluation stages of allocating tasks, setting criteria, sampling performance and/or outputs, and appraising; (2) by examining all stages in the evaluation process, and (3) by gathering information from both superintendents and principals. A questionnaire sent to a random sample of 200 principals and to all 142 superintendents in a southern state yielded 149 usable returns from principals (74%) and 121 usable returns from superintendents (85%). Results showed that (1) the evaluation system is perceived to be specific and clear in task allocations; (2) a level of uncertainty exists regarding criteria setting; and (3) principals and superintendents disagree on how performance and outputs are sampled. More adequate communication between superintendents and principals is needed. Appended are two pages of references, two tables, and one figure. (IW)