The validity and dependability of functional competency tests for adults are examined as they relate to the information needs of instructional decision makers. Test data from the Adult Performance Level (APL) Program (funded by the U.S. Office of Education at the University of Texas at Austin) is used to illustrate key points. In the discussion of validity, the importance of a test's demonstrated relevance to functional competency is discussed in terms of the definitions of the competency. Issues of content vs. criterion validity are examined particularly with reference to the APL study. Some of the problems inherent in setting and applying cutoffs (points on a scale of scores which define levels of competence) are then discussed, and the author reviews several procedures to aid in setting and adjusting cutoffs (those used by Nedelsky and by Emrick, and Bayesian techniques used by Northcutt). In the discussion of dependability (the degree to which scores are replicable) the author reviews briefly the work of Bob Brennan and Mike Kane (based on that of Cronbach and others) in the area of defining and assessing psychometric properties of criterion-referenced tests. In conclusion it is pointed out that the instructional decision maker may raise or lower a cutoff as information justifies such action but that there will be instances in which trade-offs between dependability and validity may become necessary. (JT)