The validity, as one of the criteria of 'goodness' of evaluation, is related to the very purpose of evaluation, and any attempt at reinterpreting the concept of validity involves redefining the nature of evaluation itself. In the existing theory of educational evaluation, the content validity is defined as the degree to which a test item warrants the inference to the student's knowledge of the content or his mastery of the objectives. This definition of validity entails, and is entailed by, the view which defines evaluation as the assessment of the degree to which the student has attained the instructional objectives. However, evaluation can be defined, with equal acceptance, as 'the act of empirically ascertaining the behavioral implication of the fact that the student has learned a unit of content.' This definition gives rise to an interpretation of validity, alternative to the one generally accepted in the existing theory of educational evaluation. According to this interpretation, a test item is valid if only the wrong answer to the item warrants the inference to the student's failure of learning. This interpretation is obtained by following the strict meaning of 'implication', used in the above definition of evaluation. One important implication of this new interpretation is that, by severing the intimate bond between instruction and evaluation, it proposes an alternative model of curriculum and instruction to the Tyler's. In the new interpretation, evaluation can be used, unlike in the Tylerian model, only as an indirect indicator of student's learning. It is to be noted that any attempt to 'normalize' the instructional process by improving on the validity of evaluation is doomed to be fruitless.