Performance standards are useful tools when comparative evaluation or personal selection are not appropriate for decision making. The concept was developed in 1962 when Glaser introduced criterion referenced testing, to emphasize the importance of making test scores informative about learning processes rather than comparing them to other people's scores. It was hoped that criterion referenced testing, with performance standards, would provide a simple, inexpensive decision making tool. There are three methods of setting performance standards that might allow us to relate tests to learning procedures: standards based on theories, on expert consensus, and on minimum competencies. Learning hierarchies and other theories are too simplistic; expert standard setters lack the political thrust necessary to support decisive action. The minimum competency concept of performance standards is inadequate because no single skill, whether a prerequisiste for learning or for daily existence, is so essential that it can be defined as necessary for survival. No performance standard technology exists or is likely to appear, despite the growing amount of legislation depending on such a technology. Educational technologists should search for other methods to help decision makers and inform the lay public. Criterion referenced testing has not satisfied accountability needs; partly because performance standards are so arbitrary. (CP)