New developments in measurement, especially in validity theory, in concert with new developments in cognitive and computer science afford both new reasons and new possibilities for developing direct measures of student performance. Performance processes would be assessed directly by means of work samples or simulations of real-world generic tasks, rather than in terms of total scores summarizing the piecemeal information provided by a set of discrete test items. Successful implementation of such direct assessment might not only yield more valid prediction, but also more motivationally engaging and more face-relevant predictor tests. More importantly, the approach promises to yield differentiated criterion measures that are not only more validly reflective of real-world task processes, but are also more conceptually accessible by virtue of being directly interpretable in terms of the goals or desired outcomes of education or the workplace. More important still, in the context of educational or training programs-whether in college or on the job, whether in medical school or in graduate.