Abstract. The purpose of this research was to examine frame-of-reference (FOR) training retention in an assessment center (AC) rater training context. In this study, we extended Gorman and Rentsch’s (2009) research showing FOR training effects on performance schemas by examining the effects immediately after training and again after a two-week nonuse period. We examined the retention effects of FOR training on performance ratings and on performance schema accuracy. The results indicated that the FOR training condition, compared to a control condition, yielded performance ratings and performance schemas more similar to expert ratings and to an expert schema, respectively. FOR training also had positive effects on ratings and performance schema accuracy assessed two weeks after training. These results support and extend the theory of FOR training, which posits that the instructed theory of performance replaces the preexisting rater schemas ( Lievens, 2001 ), and they contribute to the research on FOR training within AC contexts.