Using the release from proactive interference (RPI) task, college students (n = 40) received 4 trials comprised of 3-word triads of either positive (P) or negative (N) words. Word-triad recall served as the dependent measure, and results revealed typical buildup of PI (i.e., no significant group x trial interaction across Trials 1-3). Significant RPI was also observed when examining performance across Trials 3-4, and PI release was greater in those subjects who experienced a valence shift from P words to N words or from N words to P words across Trials 3-4. No release was observed in the control group and the word-triad recall performance of the control group continued to decline across trials. Results are discussed in terms of using the RPI paradigm with emotionally valenced stimuli and including clinically relevant samples.