This paper establishes and explains a series of propositions related to the installation of an evaluation system in an educational setting. The propositions are drawn from the problems identified by the Phi Delta Kappa National Study Committee on Evaluation. The paper treats five key problems and selected strategies for their recognition and avoidance. The problems are those of defining evaluation, identifying decision-makers, eliciting the values and criteria on which decisions are based, serving multiple decision levels, and distinguishing between research and evaluation. (Author)