Three experiments were conducted to examine false recall and recognition with lists of phonologically related words. Experiment 1 found that the pattern of false memories (both recall and recognition) obtained for lists of phonological associates was similar to results that have been observed with semantically associated word lists (Roediger & McDermott, 1995). Experiment 2 demonstrated that increasing the number of talkers producing list items did not significantly affect either false recall or recognition. Experiment 2 also showed that the representations underlying false memories can contain highly detailed voice information. Experiment 3 found that changing to lists with the least, rather than most, confusable phonological associates of critical (nonpresented) items significantly reduced the incidence of false memories. The findings are discussed within the framework of the Neighborhood Activation Model of spoken word recognition and suggest that similar mechanisms may mediate false memories with lists of semantic and phonological associates.