The value of research on the accuracy of job analysis is questioned. It is argued that the traditional criteria employed to evaluate job analysis accuracy (i.e., interrater agreement and deviations from proxy true scores) provide information of little practical value. Alternative criteria focusing on the consequences of job analysis data are suggested. Consequence‐oriented criteria are clarified through a review of the various inferential leaps or decision points that job analysis supports. In addition, the consequences of job analysis are also thought to be a function of the rules governing the making of job‐analysis‐based inferences which, unfortunately, are sometimes unspecified in even the most molecular job analysis methodologies. Copyright © 2000 John Wiley & Sons, Ltd.