There is no evidence to support the belief that training raters to change rating distributions will increase accuracy or validity. Such training may merely promote a temporary and situation-specific response set. We call for a new emphasis in rater training programs on: (1) diary-keeping procedures to increase observational skills; (2) the establishment of a common rater frame of reference to enhance agreement on what constitutes effective job performance; and (3) mastery-based training to increase rater self-efficacy regarding negative appraisal situations.