Although predictive validation is the primary method of justifying a particular work sample for use as an assessment device, content validation is being recognized as an alternative strategy. Confusion exists over whether content validity should refer to the representativeness of the work sample, in terms of tasks to be performed, or behavioral aspects of work performance. The in-basket exercise, a simulation of the typical contents of a manager's in-basket which require decision and action, is the work sample used with 18 lower- and middle-level managers. Effects were investigated of two in-basket exercises based upon task sampling one of which was job-related to the subjects, and two of which used job-related criteria. Results suggest that job-relatedness of assessment criteria is more important than job-relatedness of the in-basket exercise, implying a need to redefine content validity. Content validity is implied to relate to behavioral aspects of work performance rather than to the task content. (Author/NRB)