The purpose of this study was to investigate whether prior performance variations within and among job candidates affect evaluations of present performance and whether these variations result in ratings that are exaggerated or erroneous. There were three conditions: a consistent performance condition (CP), a within candidate performance variation condition (WCV), and a between candidate performance variation condition (BCV). Contrast effects were found in both the BCV and WCV conditions. In addition, ratings obtained when there were performance variations within and among candidates were significantly more accurate than those obtained when there were no performance variations. Practical implications and future research suggestions are discussed.