This article: deals with the limitations on the use of work samples as predidtors of vocational success. The predictive model is briefly reviewed and ways in which the assumptions underlying this model can be violated or altered in vocational evaluation settings are described. Simple interventions have been shown to substantially alter work sample scores .Of particular concern are those variations which vocational evaluators themselves introduces into the evaluation situation.The potential for abuse of work sample evaluation is great and, in the absence of knowledge and respect for the available technology, is likely.It is an article of faith among vocational evaluators that work samples are predictors of vocational success. There are even some who believe that work samples are such superlative predictors that we could do away with all other forms of predictors, such as norm referenced tests.The key words in that statement are &dquo;faith&dquo; and &dquo;belief.&dquo; If you like a world of honest evidence-one which has its roots in pure, simple, credible, empirical fact-you'll search far and wide and unsuccessfully for evidence to back up the faith and belief which evaluators have in work samples as predictors.A recent review of the research evidence (Dunn, 1976c) reached the following conclusions. First, our ability to predict the future job performance of anyone by any means is quite limited. Second, the evidence is that the future job performance of the disabled is even less predictable than that of the non-disabled. Third, the limited evidence of prediction which is available for vocational evaluation has been obtained in relation to rather gross criteria (such as unemployed-employed); the ability of vocational evaluation procedures to differentially predict individual success between occupations has not been demonstrated. Fourth, the requirements of a true predictive validity study for vocational evaluation, in terms of time, cost, and subjects, are such that it is unlikely that one could be done. The overall conclusion of this review was that evaluators may have to settle for &dquo;presumptive validity.&dquo; This is a type of validiat UNIV CALIFORNIA SANTA BARBARA on June 30, 2015 aei.sagepub.com Downloaded from