DNA evidence is one of the most significant modern advances in the search for truth since the cross examination, but its format as a random-match-probability makes it difficult for people to assign an appropriate probative value [1]. While Frequentist theories propose that the presentation of the match as a frequency rather than a probability facilitates more accurate assessment [2], Exemplar-Cueing Theory predicts that the subjective weight assigned may be affected by the frequency or probability format, and how easily examples of the event, i.e., ‘exemplars’, are generated from linguistic cues that frame the match in light of further evidence [3]. This paper presents two juror research studies to examine the difficulties that jurors have in assigning appropriate probative value to DNA evidence when contradictory evidence is presented. Study 1 showed that refuting evidence significantly reduced guilt judgments when exemplars were linguistically cued, even when the probability match and the refuting evidence had the same objective probative value. Moreover, qualitative reason for judgment responses revealed that interpreting refuting evidence was found to be complex and not necessarily reductive; refutation was found indicative of innocence or guilt depending on whether exemplars have been cued or not. Study 2 showed that the introduction of judges’ directions to linguistically cue exemplars, did not increase the impact of refuting evidence beyond its objective probative value, but less guilty verdicts were returned when jurors were instructed to consider all possible explanations of the evidence. The results are discussed in light of contradictory frequentist and exemplar-cueing theoretical positions, and their real-world consequences.